Corrections & updates
Every material factual change to a published page is recorded here: what the page said, what it says now, and why. A correction means the page was wrong when it was published. An update means the facts changed afterwards and the page was re-verified. The rules behind this log are in the editorial policy.
253 entries: 124 corrections, 129 updates. Found an error? [email protected].
November 18, 2026 — Update (1 page)
Updated /answers/what-is-gpt-5-6-sol-terra-luna-openai-june-2026/
- Was: OpenAI's current generation is GPT-6 (GPT-6 Astra, GPT-6.1 Sol, GPT-6.1 Terra, GPT-6 Luna); GPT-6 Astra flagship at $10/$50
- Now: Current GPT-6 lineup includes GPT-6.2 Astra ($8/$40) and GPT-6.2 Sol ($2/$10), released November 17, 2026, plus GPT-6.1 Terra and GPT-6 Luna
- Why it happened: claim audit
- Source: https://openai.com/index/introducing-gpt-6-2-astra-sol/
- Page: /answers/what-is-gpt-5-6-sol-terra-luna-openai-june-2026/
November 18, 2026 — Correction (1 page)
Corrected /answers/what-is-gpt-5-6-sol-terra-luna-openai-june-2026/
- Was: GPT-5.5 83.4% on Terminal-Bench 2.1
- Now: GPT-5.5 85.6% on Terminal-Bench 2.1
- Why it happened: claim audit
- Source: https://openai.com/index/gpt-5-6/
- Page: /answers/what-is-gpt-5-6-sol-terra-luna-openai-june-2026/
November 18, 2026 — Correction (1 page)
Corrected /answers/what-is-qwen4-max-price-specs-access-2026/
- Was: Model Studio's Chinese mainland page marks batch inference and fine-tuning as not supported
- Now: Model Studio marks batch inference as supported in the Chinese mainland scope (not in the International and Global scopes) and fine-tuning as not supported
- Why it happened: claim audit
- Source: https://docs.modelstudio.console.alibabacloud.com/en/model-studio/qwen4-max
- Page: /answers/what-is-qwen4-max-price-specs-access-2026/
November 18, 2026 — Correction (1 page)
Corrected /answers/best-mac-for-running-local-llms-2026/
- Was: M6 153GB/s
- Now: M6 153GB/s (16GB) or 170GB/s (24GB or 32GB)
- Why it happened: claim audit
- Source: https://www.apple.com/macbook-pro/specs/
- Page: /answers/best-mac-for-running-local-llms-2026/
November 17, 2026 — Update (1 page)
Updated /answers/what-is-gpt-5-6-sol-terra-luna-openai-june-2026/
- Was: Sol $5/$30, Terra $2.50/$15, Luna $1/$6 per 1M tokens
- Now: Sol $4/$20 (promotional, until Nov 30, 2026), Terra $2/$12, Luna $0.20/$1.20 per 1M tokens
- Source: https://developers.openai.com/api/docs/pricing
- Page: /answers/what-is-gpt-5-6-sol-terra-luna-openai-june-2026/
November 17, 2026 — Update (1 page)
Updated /answers/what-is-gpt-5-6-sol-terra-luna-openai-june-2026/
- Was: Access limited to vetted partners; broader rollout expected within weeks
- Now: Generally available in ChatGPT, Codex and the OpenAI API since July 9, 2026; succeeded by the GPT-6 family
- Source: https://openai.com/index/gpt-5-6/
- Page: /answers/what-is-gpt-5-6-sol-terra-luna-openai-june-2026/
November 17, 2026 — Update (1 page)
Updated /answers/jetbrains-air-vs-cursor-3-vs-claude-code-may-2026/
- Was: Cursor Pro $20/mo; JetBrains Air released through Q1 2026; Claude Code built around Claude Opus 4.7
- Now: New Cursor customers buy The X Subscription (Plus $8, Premium $30, Super $100, Ultra $200; third-party models at API list price); Air in JetBrains IDEs GA with the 2026.3 release (Nov 2026); Claude Code on Opus 5.5 / Fable 5.2
- Source: https://blog.jetbrains.com/ai/2026/11/air-in-jetbrains-ides-ga/
- Page: /answers/jetbrains-air-vs-cursor-3-vs-claude-code-may-2026/
November 17, 2026 — Correction (1 page)
Corrected /answers/jetbrains-air-vs-cursor-3-vs-claude-code-may-2026/
- Was: JetBrains AI Pro plans starting around $24/mo per user
- Now: JetBrains AI Pro €10/month, AI Ultimate €30/month (JetBrains store, EU, Nov 17, 2026); Air in JetBrains IDEs needs no paid JetBrains AI plan
- Source: https://www.jetbrains.com/ai-ides/buy/
- Page: /answers/jetbrains-air-vs-cursor-3-vs-claude-code-may-2026/
November 17, 2026 — Update (1 page)
Updated /answers/chatgpt-work-vs-claude-cowork-vs-gemini-agents-august-2026/
- Was: Gemini agents on Gemini 3.6 Flash, with Gemini 3.5 Pro expected mid-August 2026
- Now: Gemini 3.5 Pro never reached Google's API pricing page; Google announced the Gemini agent on Oct 8, 2026 and runs it on the Gemini 4 models (Argon, Carbon, Hydrogen)
- Source: https://cloud.google.com/blog/products/ai-machine-learning/welcome-to-gemini-at-work-2026
- Page: /answers/chatgpt-work-vs-claude-cowork-vs-gemini-agents-august-2026/
November 17, 2026 — Update (1 page)
Updated /answers/chatgpt-work-vs-claude-cowork-vs-gemini-agents-august-2026/
- Was: ChatGPT Work runs on GPT-5.6; Claude Cowork on Claude Opus 5, which leads Artificial Analysis's indexes
- Now: ChatGPT Work runs on GPT-6.1 Terra (since Oct 26, 2026), GPT-6.1 Sol and GPT-6 Astra; Claude Cowork on Opus 5.5 / Fable 5.2, merged into the main Claude app and running in Anthropic's cloud since Oct 6
- Source: https://openai.com/index/introducing-gpt-6-1-terra/
- Page: /answers/chatgpt-work-vs-claude-cowork-vs-gemini-agents-august-2026/
November 17, 2026 — Update (1 page)
Updated /answers/best-agentic-coding-frameworks-2026/
- Was: Superpowers 124K, BMAD ~5K, SpecKit ~3K, OpenSpec ~2K stars; 170K+ combined
- Now: Superpowers 319K, Spec Kit 150K, OpenSpec 75K, BMAD 55K, GSD Core 12K stars (Nov 17, 2026); 600K+ combined
- Source: https://github.com/obra/superpowers
- Page: /answers/best-agentic-coding-frameworks-2026/
November 17, 2026 — Update (1 page)
Updated /answers/best-agentic-coding-frameworks-2026/
- Was: GSD (~8K stars) installed from gsd-build/get-shit-done
- Now: Original GSD repo archived (last release v1.42.3, May 16, 2026); development continues as GSD Core (open-gsd/gsd-core, v1.21.0)
- Source: https://github.com/open-gsd/gsd-core
- Page: /answers/best-agentic-coding-frameworks-2026/
November 17, 2026 — Update (2 pages)
Infisical's pricing page now lists Agent Vault overage at $1 per 10,000 brokered requests.
- Was: Infisical Agent Vault: 5,000 brokered requests a month free, then $2 per 1,000
- Now: Infisical Agent Vault: 5,000 brokered requests a month free, then $1 per 10,000
- Source: https://infisical.com/pricing
- Pages: /answers/best-secrets-managers-for-ai-agents-2026/, /answers/best-tools-for-brokering-secrets-to-ai-agents-at-runtime-2026/
November 16, 2026 — Correction (1 page)
Corrected /answers/document-data-extraction-companies-250-1000-employees-2026/
- Was: Mistral OCR 4.1: 170 languages
- Now: Mistral OCR 4.1: 40+ languages (Mistral's docs: performs strongly across more than 40 languages)
- Why it happened: claim audit
- Source: https://docs.mistral.ai/capabilities/document_ai/basic_ocr
- Page: /answers/document-data-extraction-companies-250-1000-employees-2026/
November 15, 2026 — Update (1 page)
Updated /posts/kividb-review-rust-redis-alternative-vector-search/
- Was: Open cluster and persistence edge cases among the 71 open issues (#517, #524, #545)
- Now: The three issues open at review time (#517, #524, #545) were closed as completed between November 12 and 14, 2026; 56 issues open on November 15
- Why it happened: claim audit
- Source: https://github.com/kividbio/kividb
- Page: /posts/kividb-review-rust-redis-alternative-vector-search/
November 14, 2026 — Correction (1 page)
Corrected /answers/best-enterprise-ai-ocr-document-processing-vendors-2026/
- Was: LlamaParse at $1.25 per 1,000 credits
- Now: LlamaParse at $1.50 per 1,000 credits pay-as-you-go ($1.25 on the $500/month Pro plan)
- Why it happened: claim audit
- Source: https://www.llamaindex.ai/pricing
- Page: /answers/best-enterprise-ai-ocr-document-processing-vendors-2026/
November 14, 2026 — Correction (1 page)
Corrected /posts/kividb-review-rust-redis-alternative-vector-search/
- Was: 106 forks (facts as of November 10, 2026)
- Now: 115 forks (facts as of November 10, 2026: GitHub's fork list shows 115 forks created before that morning's snapshot, whose 71 open / 210 closed issue counts match exactly)
- Why it happened: claim audit
- Source: https://github.com/kividbio/kividb
- Page: /posts/kividb-review-rust-redis-alternative-vector-search/
November 14, 2026 — Update (1 page)
Updated /answers/alternatives-to-dosu-ai-memory-knowledge-systems-2026/
- Was: Mem0: Hobby free, Starter $19, Pro $249 a month; Pro adds graph memory and Dream, Enterprise on-prem
- Now: Mem0: Free tier (10,000 adds / 1,000 retrievals a month for 30 days; 50,000 / 10,000 with your own LLM key), then custom plans with graph memory, Dream and on-prem; Starter, Growth and Pro are no longer sold to new accounts
- Why it happened: claim audit
- Source: https://mem0.ai/pricing
- Page: /answers/alternatives-to-dosu-ai-memory-knowledge-systems-2026/
November 13, 2026 — Update (1 page)
Updated /answers/trusted-agent-control-plane-providers-enterprise-2026/
- Was: Google Agent Gateway governs agents on Google Cloud; Salesforce: only Agent Fabric and existing Agentforce tools until FY28
- Now: Agent Gateway for cross-cloud in Limited Preview since Nov 11, 2026 (EKS, AKS, other CNCF Kubernetes); Salesforce Agent Fabric added hard spending limits and an agent kill switch on Oct 27, 2026
- Source: https://docs.cloud.google.com/gemini-enterprise-agent-platform/release-notes
- Page: /answers/trusted-agent-control-plane-providers-enterprise-2026/
November 13, 2026 — Update (1 page)
Updated /answers/best-enterprise-ai-ocr-document-processing-vendors-2026/
- Was: Reducto: Parse $10 per 1,000 pages (cheapest Reducto parse option)
- Now: Reducto r-1 mini Parse $1 per 1,000 pages since Nov 5, 2026; r-1 Parse still $10, Extract $20, Deep Extract $40
- Source: https://reducto.ai/pricing
- Page: /answers/best-enterprise-ai-ocr-document-processing-vendors-2026/
November 13, 2026 — Update (1 page)
Updated /answers/ios-27-liquid-glass-vs-ios-26-accessibility-changes-june-2026/
- Was: iOS 27.2 is in beta; install iOS 27.0.1 or later
- Now: iOS 27.2 released Oct 26, 2026 and iOS 27.2.1 on Nov 9, 2026; iOS 27.3 in beta since Nov 3
- Source: https://www.macworld.com/article/2986799/ios-27-features-compatiblity-siri-ai-updates.html
- Page: /answers/ios-27-liquid-glass-vs-ios-26-accessibility-changes-june-2026/
November 13, 2026 — Update (1 page)
Updated /answers/best-open-weight-image-generation-models-2026-ranked/
- Was: Six ranked open-weight image models; newest Qwen-Image-2.1 (Sep 20, 2026)
- Now: Added FLUX 3 Image Klein (Black Forest Labs, Oct 15, 2026, 5.6B, FLUX Kommunity License: non-commercial plus commercial output under US$5M revenue) as a new unranked entry, and Qwen-Image-2.1-Turbo (8-step, Oct 9, 2026)
- Source: https://huggingface.co/black-forest-labs/flux-3-image-klein
- Page: /answers/best-open-weight-image-generation-models-2026-ranked/
November 13, 2026 — Update (1 page)
Updated /answers/should-you-buy-anthropic-ipo-shares-june-2026/
- Was: SoFi, Robinhood, and Webull may offer IPO Access participation for retail
- Now: S-1/A names Schwab, Fidelity, Robinhood, Trade-PMR, SoFi and E*TRADE as retail selling-group members; Webull is not listed
- Source: https://www.sec.gov/Archives/edgar/data/2130357/000162828026072484/anthropicpbc-sx1ax2.htm
- Page: /answers/should-you-buy-anthropic-ipo-shares-june-2026/
November 13, 2026 — Update (1 page)
Updated /answers/should-you-buy-anthropic-ipo-shares-june-2026/
- Was: Standard 180-day lock-up; ~3.5B insider shares eligible to sell after 180 days
- Now: Staged lock-up: up to 316M shares tradable from day one, tranches on days 48/62/76 and after FY2026 and Q1 2027 earnings, 180 days max for most holders, up to 365 days for co-founders and >0.5% holders
- Source: https://www.sec.gov/Archives/edgar/data/2130357/000162828026072484/anthropicpbc-sx1ax2.htm
- Page: /answers/should-you-buy-anthropic-ipo-shares-june-2026/
November 13, 2026 — Update (1 page)
Updated /answers/should-you-buy-anthropic-ipo-shares-june-2026/
- Was: Expected debut October-November 2026; likely market cap $850B-$1.2T
- Now: S-1/A (Nov 10, 2026): $70-$78 per share, 825M shares, ticker ANT; pricing expected Nov 18 and trading Nov 19, 2026; about $1.4T fully diluted at the top of the range
- Source: https://www.sec.gov/Archives/edgar/data/2130357/000162828026072484/anthropicpbc-sx1ax2.htm
- Page: /answers/should-you-buy-anthropic-ipo-shares-june-2026/
November 13, 2026 — Update (1 page)
Added Moxie Docs, a GitHub-based documentation-maintenance tool, as a further Dosu alternative
- Was: Five tools compared: Dosu, DeepWiki, Context7, Unblocked and Mem0
- Now: Moxie Docs added: an entry, a column in the comparison table and a line in How to choose, from its public pages (November 13, 2026)
- Source: https://moxiedocs.com/pricing
- Page: /answers/alternatives-to-dosu-ai-memory-knowledge-systems-2026/
November 13, 2026 — Update (1 page)
Updated /posts/kividb-review-rust-redis-alternative-vector-search/
- Was: Latest release v1.2.0 (November 3, 2026)
- Now: Latest release v1.2.2 (November 12, 2026), fixing advisory GHSA-8rm8-p5c5-g3mr in 1.2.0 and 1.2.1; v1.2.1 the same day fixed 1.2.0's reshard and replica known issues
- Why it happened: claim audit
- Source: https://github.com/kividbio/kividb/releases/tag/v1.2.2
- Page: /posts/kividb-review-rust-redis-alternative-vector-search/
November 13, 2026 — Update (1 page)
Updated /posts/gauge-cli-review-agent-experience-testing/
- Was: Latest release CLI 0.29.0 (November 9, 2026); changelog lists 27 versions
- Now: Latest release CLI 0.30.0 (November 12, 2026); changelog lists 28 versions
- Why it happened: claim audit
- Source: https://github.com/gauge-sh/gauge-cli/blob/main/cli/CHANGELOG.md
- Page: /posts/gauge-cli-review-agent-experience-testing/
November 13, 2026 — Correction (1 page)
Corrected /answers/knowledge-layer-for-large-language-models-tools-2026/
- Was: Cloudflare AI Search billing from November 1, 2026
- Now: Cloudflare AI Search billing from December 1, 2026
- Why it happened: claim audit
- Source: https://developers.cloudflare.com/ai-search/platform/limits-pricing/
- Page: /answers/knowledge-layer-for-large-language-models-tools-2026/
November 12, 2026 — Update (1 page)
Updated /posts/caveman-claude-code-skill-token-savings-review/
- Was: Latest v4.0.1 (October 13, 2026); installer pinned to v4.0.1
- Now: Latest v4.1.0 (November 8, 2026); installer pinned to v4.1.0
- Source: https://github.com/JuliusBrussee/caveman/releases/tag/v4.1.0
- Page: /posts/caveman-claude-code-skill-token-savings-review/
November 12, 2026 — Update (1 page)
Updated /posts/alibaba-open-code-review-cli-review/
- Was: Latest release v1.12.12 (October 5, 2026)
- Now: Latest release v1.13.8 (November 12, 2026); --tools CLI overrides removed
- Source: https://github.com/alibaba/open-code-review/releases/tag/v1.13.8
- Page: /posts/alibaba-open-code-review-cli-review/
November 12, 2026 — Update (1 page)
Updated /posts/browser-use-ai-agent-browser-automation/
- Was: ChatBrowserUse bu-2-0: $0.60 input / $0.06 cached / $3.50 output per 1M tokens
- Now: 1.0 default bu-fast: $0.12 input / $0.012 cached / $0.60 output per 1M tokens (bu-2-0 only on browser-use<1.0)
- Source: https://docs.browser-use.com/open-source/models.md
- Page: /posts/browser-use-ai-agent-browser-automation/
November 12, 2026 — Update (1 page)
Updated /posts/browser-use-ai-agent-browser-automation/
- Was: Latest release 0.13.11; Browser() launches a local browser; Browser.from_system_chrome(); ChatOllama class; @sandbox() decorator
- Now: Latest release 1.0.2 (rewrite shipped Nov 10, 2026); Browser() is a cloud browser, Browser('chromium') local, Browser('chrome') attaches to your Chrome; ChatOllama and @sandbox() removed
- Source: https://github.com/browser-use/browser-use/blob/main/MIGRATING.md
- Page: /posts/browser-use-ai-agent-browser-automation/
November 12, 2026 — Update (1 page)
Updated /posts/gpt-researcher-deep-research-agent-review/
- Was: No version or Python requirement stated; last push July 2026
- Now: Latest release v3.8.0 (Nov 9, 2026; pip 0.17.0); Python 3.12+ required since v3.7.0; default writer model now openai:gpt-6-sol
- Source: https://github.com/assafelovic/gpt-researcher/releases/tag/v3.8.0
- Page: /posts/gpt-researcher-deep-research-agent-review/
November 12, 2026 — Correction (1 page)
Mem0 plans updated
- Was: Mem0: Hobby free, Starter $19, Pro $249
- Now: Mem0 pricing page lists Free and Custom plans only; graph memory on Custom
- Why it happened: vendor changed its plans after publication
- Source: https://mem0.ai/pricing
- Page: /answers/knowledge-layer-for-large-language-models-tools-2026/
November 12, 2026 — Update (1 page)
Gemini 4 Carbon availability and price schedule updated
- Was: Gemini 4 Carbon not yet generally available; introductory price end date not stated
- Now: Available to all paying Gemini API developers, Gemini for business and AI Ultra since Nov 10, 2026; $2/$10 through Nov 30, $4/$20 from Dec 1, 2026
- Source: https://aistudio.google.com/docs/pricing
- Page: /answers/what-is-gemini-4-carbon-google-frontier-model-2026/
November 12, 2026 — Correction (1 page)
Corrected /posts/kividb-review-rust-redis-alternative-vector-search/
- Was: CI runs Redis 8.10.2's own test suite on every change: 2,697 tests pass and 6 fail
- Now: CI has run Redis 8.10.2's suite since 1.1.1; the latest published figures (Redis 8.0.0's suite) are 1,892 passed, 31 failed
- Why it happened: claim audit
- Source: https://github.com/kividbio/kividb/blob/master/docs/REDIS_TEST_SUITE.md
- Page: /posts/kividb-review-rust-redis-alternative-vector-search/
November 12, 2026 — Correction (1 page)
Corrected /answers/best-open-source-vision-language-models-july-2026/
- Was: MolmoMotion released July 1, 2026
- Now: MolmoMotion released June 17, 2026
- Why it happened: claim audit
- Source: https://allenai.org/blog/molmo-motion
- Page: /answers/best-open-source-vision-language-models-july-2026/
November 12, 2026 — Correction (1 page)
Corrected /answers/best-ai-qa-testing-agents-2026/
- Was: The latest Playwright release is v1.64.1 (October 14, 2026)
- Now: The latest Playwright release is v1.65.0 (November 5, 2026)
- Why it happened: claim audit
- Source: https://github.com/microsoft/playwright/releases/tag/v1.65.0
- Page: /answers/best-ai-qa-testing-agents-2026/
November 12, 2026 — Correction (1 page)
Corrected /posts/kividb-review-rust-redis-alternative-vector-search/
- Was: v1.1.3, a week earlier, was a security release
- Now: v1.1.3, three days earlier (October 31), was a security release
- Why it happened: claim audit
- Source: https://github.com/kividbio/kividb/releases/tag/v1.1.3
- Page: /posts/kividb-review-rust-redis-alternative-vector-search/
November 11, 2026 — Update (1 page)
Added TestOptim AI (web-app QA without test code) from the maker's public pages
- Was: Four tools compared: Momentic Mo, QA Wolf, mabl, Playwright Test Agents
- Now: TestOptim AI added: a no-test-code option for web apps, with prices from its pricing page
- Source: https://www.testoptim.com/pricing
- Page: /answers/best-ai-qa-testing-agents-2026/
November 11, 2026 — Update (1 page)
Added Humla's language support from its own site and README
- Was: Humla entry did not state its language support
- Now: Humla entry states it handles around 100 languages (everything Whisper supports) with auto-detection and per-language engine routing, e.g. a local Norwegian-tuned Whisper model
- Source: https://humla.team/
- Page: /answers/best-local-ai-meeting-note-takers-2026-foresight-anarlog-granola/
November 11, 2026 — Correction (1 page)
Corrected /answers/best-open-source-vision-language-models-july-2026/
- Was: Qwen3-VL strengths: ... 100+ languages
- Now: Qwen3-VL strengths: ... OCR in 32 languages
- Why it happened: claim audit
- Source: https://huggingface.co/Qwen/Qwen3-VL-8B-Instruct
- Page: /answers/best-open-source-vision-language-models-july-2026/
November 11, 2026 — Update (1 page)
Updated /posts/cloudflare-security-audit-skill-review-agent-vulnerability-hunting/
- Was: 211-point Hacker News thread
- Now: 213-point Hacker News thread
- Why it happened: claim audit
- Source: https://news.ycombinator.com/item?id=49736466
- Page: /posts/cloudflare-security-audit-skill-review-agent-vulnerability-hunting/
November 11, 2026 — Update (1 page)
Updated /posts/kividb-review-rust-redis-alternative-vector-search/
- Was: The README's headline (21 million pipelined GETs/s, about 9× Redis 8 and 2.3× Dragonfly) comes from an October bare-metal session with KiviDB 1.1.0
- Now: Since November 10 the README's headline (21.2 million pipelined GETs/s, 9× Redis 8.10, 2.3× Dragonfly, ahead of Garnet) comes from a November bare-metal run of KiviDB 1.2.0; the October 1.1 run is kept in benchmarks/1.1-bare-metal.md
- Why it happened: claim audit (README changed after publication)
- Source: https://github.com/kividbio/kividb
- Page: /posts/kividb-review-rust-redis-alternative-vector-search/
November 10, 2026 — Correction (1 page)
Corrected /answers/best-open-weight-coding-model-2026-glm-deepseek-qwen-guide/
- Was: Single-GPU pick: Qwen3.6-27B
- Now: Single-GPU pick: Qwen3.8-27B (released Aug 5, 2026; 73.0 vs 63.4 Terminal-Bench 2.1 in Qwen's table)
- Source: https://huggingface.co/Qwen/Qwen3.8-27B
- Page: /answers/best-open-weight-coding-model-2026-glm-deepseek-qwen-guide/
November 10, 2026 — Update (1 page)
Updated /answers/best-open-weight-coding-model-2026-glm-deepseek-qwen-guide/
- Was: Overall agentic pick: GLM-5.2 (81.0 Terminal-Bench 2.1)
- Now: GLM-5.3 (88.2 Terminal-Bench 2.1) and MiMo-V2.6-Pro (89.9) lead; Reflection Beam 1.0 added
- Source: https://huggingface.co/zai-org/GLM-5.3
- Page: /answers/best-open-weight-coding-model-2026-glm-deepseek-qwen-guide/
November 10, 2026 — Update (1 page)
Updated /answers/best-open-weight-coding-model-2026-glm-deepseek-qwen-guide/
- Was: DeepSeek V4 Pro API $0.435/$0.87 per MTok off-peak
- Now: DeepSeek V4 Pro $0.66/$1.98 off-peak, $1.32/$3.96 peak (repriced Aug 16, 2026)
- Source: https://api-docs.deepseek.com/quick_start/pricing
- Page: /answers/best-open-weight-coding-model-2026-glm-deepseek-qwen-guide/
November 10, 2026 — Correction (1 page)
Corrected /answers/best-open-source-vision-language-models-july-2026/
- Was: All five models are supported by Ollama
- Now: Ollama carries qwen3-vl, qwen3.8, gemma4, llama4; not Molmo or DeepSeek VL2
- Source: https://ollama.com/search?c=vision
- Page: /answers/best-open-source-vision-language-models-july-2026/
November 10, 2026 — Correction (1 page)
Corrected /answers/best-open-source-vision-language-models-july-2026/
- Was: DeepSeek VL: MIT license, 7B and 33B
- Now: DeepSeek VL2 (Tiny/Small/base MoE) under the DeepSeek model license
- Source: https://huggingface.co/deepseek-ai/deepseek-vl2
- Page: /answers/best-open-source-vision-language-models-july-2026/
November 10, 2026 — Correction (1 page)
Corrected /answers/best-open-source-vision-language-models-july-2026/
- Was: Llama 4 Vision in 90B and 400B multimodal sizes
- Now: Llama 4 Scout (109B total, 17B active) and Maverick (400B total, 17B active); no 90B Llama 4
- Source: https://ai.meta.com/blog/llama-4-multimodal-intelligence/
- Page: /answers/best-open-source-vision-language-models-july-2026/
November 10, 2026 — Correction (1 page)
Corrected /answers/best-open-source-vision-language-models-july-2026/
- Was: Qwen 3 VL sizes 2B, 7B, 32B, 72B
- Now: Qwen3-VL sizes 2B, 4B, 8B, 32B, 30B-A3B, 235B-A22B (7B/72B are Qwen2.5-VL)
- Source: https://huggingface.co/Qwen/Qwen3-VL-8B-Instruct
- Page: /answers/best-open-source-vision-language-models-july-2026/
November 10, 2026 — Update (1 page)
Updated /answers/is-anthropic-profitable-first-profit-q2-2026-explained/
- Was: No public S-1 on EDGAR; IPO timing unconfirmed
- Now: Public S-1 filed Oct 26, 2026; Amendment No. 2 (Nov 10) sets $70-$78/share range, ticker ANT on Nasdaq
- Source: https://www.anthropic.com/news/s1-sec
- Page: /answers/is-anthropic-profitable-first-profit-q2-2026-explained/
November 10, 2026 — Update (1 page)
Updated /answers/is-anthropic-profitable-first-profit-q2-2026-explained/
- Was: Q3 2026 projected ~$1B GAAP EBITDA (SemiAnalysis); no Q3 results published
- Now: Q3 2026 reported: $16.96B revenue, $567M GAAP operating loss, $2.5B adjusted operating income (S-1/A)
- Source: https://www.sec.gov/Archives/edgar/data/2130357/000162828026072484/anthropicpbc-sx1ax2.htm
- Page: /answers/is-anthropic-profitable-first-profit-q2-2026-explained/
November 10, 2026 — Update (1 page)
Updated /answers/is-anthropic-profitable-first-profit-q2-2026-explained/
- Was: Q2 2026: ~$10.9B revenue, ~$559M operating profit (preliminary reports)
- Now: Q2 2026: $11.56B revenue, $974M operating income (Anthropic S-1)
- Source: https://www.sec.gov/Archives/edgar/data/2130357/000162828026072484/anthropicpbc-sx1ax2.htm
- Page: /answers/is-anthropic-profitable-first-profit-q2-2026-explained/
November 10, 2026 — Update (1 page)
Added Humla to the comparison
- Was: The page compared Anarlog, Google AI Edge Foresight, Kimiko, Granola, Kasset, Fireflies and Otter.
- Now: Humla, a free MIT-licensed Mac app, added: an entry, a comparison-table row and a line in the how-to-choose list, from its website and README.
- Source: https://github.com/michaelwilhelmsen/humla
- Page: /answers/best-local-ai-meeting-note-takers-2026-foresight-anarlog-granola/
November 9, 2026 — Update (1 page)
Kimi K3.1 Pro open weights are now public; page updated from 'promised' to 'released'.
- Was: Open weights due by November 4, 2026; not yet released
- Now: Weights released on Hugging Face (moonshotai/Kimi-K3.1-Pro), 2.8T MoE / 104B active, MXFP4, modified MIT licence
- Source: https://huggingface.co/moonshotai/Kimi-K3.1-Pro
- Page: /answers/what-is-kimi-k3-1-pro-moonshot-price-weights-2026/
November 9, 2026 — Correction (1 page)
Corrected /posts/craft-anything-review-rust-reverse-engineering-ai-rebuild/
- Was: 166 open issues (issues only; GitHub's combined figure of 301 also counts open pull requests)
- Now: 166 open issues (issues only; GitHub's combined open-issues figure also counts open pull requests) — the 301 combined figure could not be reproduced (166 issues + 122 PRs open at publication)
- Why it happened: claim audit
- Source: https://github.com/storytold/craft-anything
- Page: /posts/craft-anything-review-rust-reverse-engineering-ai-rebuild/
November 9, 2026 — Correction (1 page)
Corrected /answers/what-is-openai-gpt-6-cyber-daybreak-red-access-2026/
- Was: OpenAI has not published a public per-token price for GPT-6 Cyber
- Now: OpenAI's API pricing page lists GPT-6 Cyber at $20 input / $100 output per 1M tokens ($2 cached input); access limited to approved Daybreak Red customers
- Why it happened: claim audit
- Source: https://developers.openai.com/api/docs/pricing
- Page: /answers/what-is-openai-gpt-6-cyber-daybreak-red-access-2026/
November 9, 2026 — Correction (1 page)
Corrected /answers/what-is-openai-gpt-6-cyber-daybreak-red-access-2026/
- Was: GPT-6 Cyber is the first OpenAI model rated Critical in cybersecurity under its Preparedness Framework
- Now: GPT-6 Cyber, like GPT-6 Astra, is rated Critical in cybersecurity under OpenAI's Preparedness Framework
- Why it happened: claim audit
- Source: https://deploymentsafety.openai.com/gpt-6-cyber
- Page: /answers/what-is-openai-gpt-6-cyber-daybreak-red-access-2026/
November 9, 2026 — Correction (1 page)
Corrected /answers/best-platforms-lora-fine-tuning-dedicated-gpu-managed-2026/
- Was: Fireworks' docs state an on-demand (dedicated) deployment is the only supported way to serve trained models
- Now: Fireworks serves trained models on on-demand (dedicated) deployments, or on serverless for LoRAs on select base models
- Why it happened: claim audit
- Source: https://docs.fireworks.ai/fine-tuning/deploying-loras
- Page: /answers/best-platforms-lora-fine-tuning-dedicated-gpu-managed-2026/
November 9, 2026 — Correction (1 page)
Corrected /answers/best-platforms-lora-fine-tuning-dedicated-gpu-managed-2026/
- Was: Together GPU Clusters H100 $3.19 reserved 181+ days
- Now: Together GPU Clusters H100 $3.19 reserved 91–180 days (181+ days is contact-sales)
- Why it happened: claim audit
- Source: https://www.together.ai/pricing
- Page: /answers/best-platforms-lora-fine-tuning-dedicated-gpu-managed-2026/
November 8, 2026 — Correction (1 page)
Corrected /answers/best-speech-to-text-apis-2026-ranked/
- Was: Deepgram Nova-3 streaming $0.29 per hour (10,000 streaming hours: $2,900).
- Now: Deepgram Nova-3 streaming $0.46 per hour (10,000 streaming hours: $4,600).
- Why it happened: claim audit
- Source: https://deepgram.com/pricing
- Page: /answers/best-speech-to-text-apis-2026-ranked/
November 8, 2026 — Correction (1 page)
Corrected /answers/best-speech-to-text-apis-2026-ranked/
- Was: ElevenLabs Scribe v2 reports about 2.3% AA-WER on the non-streaming index.
- Now: ElevenLabs Scribe v2 reports about 2.2% AA-WER on the non-streaming index.
- Why it happened: claim audit
- Source: https://artificialanalysis.ai/speech-to-text
- Page: /answers/best-speech-to-text-apis-2026-ranked/
November 8, 2026 — Correction (1 page)
Corrected /answers/best-speech-to-text-apis-2026-ranked/
- Was: AssemblyAI's Voice Agent API bundles STT, LLM and TTS on one WebSocket at a flat $4.50 per hour.
- Now: AssemblyAI is retiring its Voice Agent API and for new builds quotes Universal-3.6 Pro Realtime at $0.45 per hour plus its LLM Gateway and Text-to-Speech API.
- Why it happened: claim audit
- Source: https://www.assemblyai.com/pricing
- Page: /answers/best-speech-to-text-apis-2026-ranked/
November 7, 2026 — Update (1 page)
Added Giggy as an additional speech-to-text API
- Was: Eight speech-to-text APIs ranked by price
- Now: Giggy added below the ranking: free Batch, Fast $0.04 and Streaming $0.08 per audio hour on prepaid credits, languages and limits, from Giggy's pricing and product pages; not ranked pending independent accuracy data
- Source: https://giggy.ai/pricing
- Page: /answers/best-speech-to-text-apis-2026-ranked/
November 7, 2026 — Correction (1 page)
Corrected /answers/what-is-tencent-hy4-release-price-weights-2026/
- Was: on Together AI: Qwen3.8 Max $2/$6, GLM-5.3 $1.40/$4.40, and Muse Spark Open $1.25/$4.25
- Now: on Together AI: GLM-5.3 $1.40/$4.40 and Muse Spark Open $1.25/$4.25 (Qwen3.8 Max removed: not listed on Together AI's pricing page)
- Why it happened: claim audit
- Source: https://www.together.ai/pricing
- Page: /answers/what-is-tencent-hy4-release-price-weights-2026/
November 7, 2026 — Correction (1 page)
Corrected /answers/what-is-anthropic-flashprove-provable-inference-2026/
- Was: released by Anthropic on November 3, 2026
- Now: released by Anthropic on November 2, 2026
- Why it happened: claim audit
- Source: https://www.anthropic.com/research/provable-inference
- Page: /answers/what-is-anthropic-flashprove-provable-inference-2026/
November 7, 2026 — Correction (1 page)
Corrected /answers/best-budget-cloud-providers-fine-tuning-open-source-models-2026/
- Was: reserved terms down to $3.19 for H100 (181+ days)
- Now: reserved terms down to $3.19 for H100 (91–180 days)
- Why it happened: claim audit
- Source: https://www.together.ai/pricing
- Page: /answers/best-budget-cloud-providers-fine-tuning-open-source-models-2026/
November 6, 2026 — Correction (1 page)
Corrected /answers/best-ai-agent-memory-systems-2026-hindsight-mem0-zep-letta-cognee/
- Was: Mem0 and Hindsight both MIT-licensed
- Now: Hindsight is MIT; Mem0, Graphiti, Letta and cognee are Apache 2.0
- Source: https://github.com/mem0ai/mem0
- Page: /answers/best-ai-agent-memory-systems-2026-hindsight-mem0-zep-letta-cognee/
November 6, 2026 — Update (1 page)
Updated /answers/best-ai-agent-memory-systems-2026-hindsight-mem0-zep-letta-cognee/
- Was: Mem0 hosted: free platform tier with usage pricing above it; cognee cloud free then $2.50 per 1M tokens
- Now: Mem0 dropped Starter/Pro (Nov 2026): Free (50K adds / 10K retrievals a month with your own LLM key) and Custom; cognee cloud free with 1M tokens, then a flat per-token rate plus $5 per extra workspace ($2.50 figure no longer confirmable)
- Source: https://mem0.ai/pricing
- Page: /answers/best-ai-agent-memory-systems-2026-hindsight-mem0-zep-letta-cognee/
November 6, 2026 — Update (1 page)
Updated /answers/gpt-6-sol-vs-gpt-6-luna-vs-gpt-6-astra-which-gpt-6-september-2026/
- Was: Three GPT-6 tiers; AA Intelligence Index Astra 53, Sol 48; Luna the cheapest mainstream model at $0.10/$0.50
- Now: GPT-6.1 Sol (Sep 29, $2/$10, $0.10 cached) and GPT-6.1 Terra (Oct 26, $1/$5) added; AA v5.0.1 at max: Astra 57, GPT-6.1 Sol 55, Terra 51, Luna 28; Claude Haiku 5.5 matches Luna at $0.10/$0.50 up to 100K-token prompts
- Source: https://developers.openai.com/api/docs/pricing
- Page: /answers/gpt-6-sol-vs-gpt-6-luna-vs-gpt-6-astra-which-gpt-6-september-2026/
November 6, 2026 — Update (1 page)
Updated /answers/amd-threadripper-halo-station-ai-workstation-september-2026/
- Was: No availability date announced
- Now: AMD product page: 'Coming in 2027' (still a prototype, no price or OEM partner)
- Source: https://www.amd.com/en/products/workstations/amd-threadripper-halo-station.html
- Page: /answers/amd-threadripper-halo-station-ai-workstation-september-2026/
November 6, 2026 — Correction (1 page)
Corrected /answers/amd-threadripper-halo-station-ai-workstation-september-2026/
- Was: AMD's July 22, 2026 deal with Anthropic described as a $5 billion MI450 agreement
- Now: A partnership for Anthropic to deploy up to 2 gigawatts of Instinct MI450-series GPUs from H1 2027, plus an AMD commitment to invest up to $5 billion in Anthropic
- Source: https://newsroom.amd.com/news/amd-anthropic-strategic-partnership/
- Page: /answers/amd-threadripper-halo-station-ai-workstation-september-2026/
November 6, 2026 — Update (1 page)
Updated /answers/how-to-get-claude-mythos-5-access-june-2026/
- Was: On Amazon Bedrock, the Cyber Verification Program is available only to customers with Enterprise Frontier Safeguards
- Now: On Bedrock, CVP is available only to organizations with a Fable 5.1 data retention exemption until Enterprise Frontier Safeguards support arrives; approved organizations can now also apply to use CVP access in managed services
- Source: https://support.claude.com/en/articles/14604842
- Page: /answers/how-to-get-claude-mythos-5-access-june-2026/
November 6, 2026 — Update (1 page)
Updated /answers/claude-opus-5-5-vs-claude-fable-5-1-vs-gpt-6-astra-september-2026/
- Was: Artificial Analysis Intelligence Index: Opus 5.5 58, Fable 5.1 53, GPT-6 Astra 53; cost per AA task at max $5.98 / $7.63 / $3.26
- Now: AA Intelligence Index v5.0.1 (Nov 6, 2026): Opus 5.5 59, GPT-6 Astra 57, Fable 5.1 49 at max (50 at xhigh); cost per AA task at max $10.04 / $6.24 / $12.99
- Source: https://artificialanalysis.ai/leaderboards/models
- Page: /answers/claude-opus-5-5-vs-claude-fable-5-1-vs-gpt-6-astra-september-2026/
November 6, 2026 — Update (1 page)
Updated /answers/best-tools-for-running-multiple-ai-coding-agents-simultaneously-2026/
- Was: Conductor listed as best overall with no note on its status
- Now: Added a note that Conductor joined Meta on Nov 4, 2026; Cloud read-only Nov 13, shut down Nov 20; local app fixes until early 2027; link to alternatives
- Source: https://www.conductor.build/announcement
- Page: /answers/best-tools-for-running-multiple-ai-coding-agents-simultaneously-2026/
November 6, 2026 — Update (1 page)
Maker renamed from TesterArmy to e2e; links moved to e2e.com
- Was: Named the maker as TesterArmy and linked the product page and docs at tester.army / e2e.tester.army
- Now: The company renamed itself e2e on November 3, 2026 (with an $8M seed round); name, product page and docs links updated to e2e.com, and the TesterArmy platform's January 15, 2027 shutdown date added
- Source: https://e2e.com/blog/seed
- Page: /posts/e2e-review-tester-army-ai-testing-framework/
November 6, 2026 — Correction (1 page)
Corrected /answers/best-web-search-apis-for-ai-agents-2026/
- Was: Exa $10 of credit (up to 2,500 Instant searches)
- Now: Exa $10 of credit (up to 2,500 Lite searches)
- Why it happened: claim audit
- Source: https://exa.ai/docs/admin/pricing
- Page: /answers/best-web-search-apis-for-ai-agents-2026/
November 6, 2026 — Correction (1 page)
Corrected /answers/inference-platforms-predictable-latency-2026/
- Was: Claude Opus 5.5's fast mode ($8/$40 per million tokens, Claude API only)
- Now: Claude Opus 5.5's fast mode ($8/$40 per million tokens, on the Claude API, Claude Models on Google Cloud and Microsoft Foundry)
- Why it happened: claim audit
- Source: https://platform.claude.com/docs/en/about-claude/pricing
- Page: /answers/inference-platforms-predictable-latency-2026/
November 6, 2026 — Correction (1 page)
Corrected /answers/atlassian-rovo-vs-microsoft-copilot-vs-agentforce-enterprise-agents-2026/
- Was: Atlassian announced the meters on September 2, 2026
- Now: Atlassian announced the meters on September 1, 2026
- Why it happened: claim audit
- Source: https://www.atlassian.com/blog/company-news/expanded-usage-based-pricing
- Page: /answers/atlassian-rovo-vs-microsoft-copilot-vs-agentforce-enterprise-agents-2026/
November 5, 2026 — Update (1 page)
Updated /posts/pieter-levels-photo-ai-solo-founder/
- Was: Portfolio generates ~$3.1M ARR (no margin figure)
- Now: Levels (Sept 23, 2026): business revenue stable at ~$200K-$250K/mo with ~94.5% profit margin
- Source: https://levels.io/passed-10m-revenue-investment-gains
- Page: /posts/pieter-levels-photo-ai-solo-founder/
November 5, 2026 — Update (1 page)
Updated /posts/polsia-1m-arr-30-days-zero-employees/
- Was: You can watch live company operations at polsia.com/live
- Now: The public live feed was taken down around October 12, 2026; polsia.com/live redirects to the homepage
- Source: https://aiolum.com/story/polsia/
- Page: /posts/polsia-1m-arr-30-days-zero-employees/
November 5, 2026 — Update (1 page)
Updated /posts/squad-multi-agent-teams-github-copilot/
- Was: Latest release v0.13.1, status alpha; Node.js 18+
- Now: Latest release v1.1.0 (October 20, 2026), stable since v1.0.0 (October 2, 2026); Node.js 22.5+
- Source: https://github.com/bradygaster/squad/blob/main/CHANGELOG.md
- Page: /posts/squad-multi-agent-teams-github-copilot/
November 5, 2026 — Correction (1 page)
Corrected /posts/hermes-agent-nous-research-self-improving-review/
- Was: 36.4K open issues (August 27, 2026)
- Now: That figure was GitHub's combined issues+PRs counter; November 5, 2026: ~18.6K open issues and ~34K open PRs
- Source: https://github.com/NousResearch/hermes-agent/issues
- Page: /posts/hermes-agent-nous-research-self-improving-review/
November 5, 2026 — Update (1 page)
Updated /posts/hermes-agent-nous-research-self-improving-review/
- Was: Six terminal backends: local, Docker, SSH, Daytona, Singularity, Modal; no official hosted version
- Now: Built-in backends: local, Docker, SSH, Modal, Daytona, Vercel Sandbox (Singularity is a plugin); releases ship to Hermes Cloud
- Source: https://github.com/NousResearch/hermes-agent
- Page: /posts/hermes-agent-nous-research-self-improving-review/
November 5, 2026 — Update (1 page)
Updated /posts/hermes-agent-nous-research-self-improving-review/
- Was: Native Windows is not supported — use WSL2
- Now: Native Windows is supported via a PowerShell installer and Microsoft Store app
- Source: https://github.com/NousResearch/hermes-agent
- Page: /posts/hermes-agent-nous-research-self-improving-review/
November 5, 2026 — Update (1 page)
Updated /posts/hermes-agent-nous-research-self-improving-review/
- Was: Current release v2026.8.19 (August 21, 2026)
- Now: Current release v0.22.7 (November 3, 2026); PyPI still 0.19.0
- Source: https://github.com/NousResearch/hermes-agent/releases/tag/v0.22.7
- Page: /posts/hermes-agent-nous-research-self-improving-review/
November 5, 2026 — Update (1 page)
Added Parallel Search API to the comparison
- Was: Five search APIs compared (Tavily, Exa, Brave, Perplexity, Firecrawl)
- Now: Parallel Search API added as a sixth option: modes, prices per 1,000 requests, free allowance and Extract pricing, from Parallel's pricing page and docs
- Source: https://parallel.ai/pricing
- Page: /answers/best-web-search-apis-for-ai-agents-2026/
November 5, 2026 — Correction (1 page)
Corrected /answers/how-to-set-hard-spend-limits-on-llm-apis-2026-guide/
- Was: workspace-scoped management keys can only manage keys in their own workspace
- Now: management keys are unscoped: one key covers administrative actions across all of the account's workspaces
- Why it happened: claim audit
- Source: https://openrouter.ai/docs/guides/overview/auth/management-api-keys
- Page: /answers/how-to-set-hard-spend-limits-on-llm-apis-2026-guide/
November 5, 2026 — Correction (1 page)
Corrected /answers/baseten-batch-processing-spot-gpu-evaluation-2026/
- Was: H100 SXM $2.69 vs $3.99/hour on October 28, 2026
- Now: H100 SXM $2.69 vs $3.99/hour on November 4, 2026
- Why it happened: claim audit
- Source: https://www.runpod.io/pricing
- Page: /answers/baseten-batch-processing-spot-gpu-evaluation-2026/
November 5, 2026 — Correction (1 page)
Corrected /answers/baseten-pricing-vs-modal-replicate-runpod-ml-infrastructure-2026/
- Was: RunPod's pricing page (updated October 28, 2026)
- Now: RunPod's pricing page (updated November 4, 2026)
- Why it happened: claim audit
- Source: https://www.runpod.io/pricing
- Page: /answers/baseten-pricing-vs-modal-replicate-runpod-ml-infrastructure-2026/
November 4, 2026 — Correction (1 page)
Corrected /answers/ios-27-siri-extensions-hidden-claude-chatgpt-gemini-june-2026/
- Was: Apple has only signed a public partnership with OpenAI so far — the existing ChatGPT-in-Siri integration that shipped in iOS 18.
- Now: Apple's only public third-party assistant partnership is with OpenAI (ChatGPT in Siri since iOS 18); separately, Apple says its Apple Foundation Models were custom-built in collaboration with Google and its Gemini models.
- Why it happened: claim audit
- Source: https://www.apple.com/newsroom/2026/09/siri-ai-a-profoundly-more-capable-and-personal-assistant-is-here/
- Page: /answers/ios-27-siri-extensions-hidden-claude-chatgpt-gemini-june-2026/
November 4, 2026 — Correction (1 page)
Corrected /answers/baseten-vs-together-ai-vs-fireworks-ai-inference-platforms-june-2026/
- Was: Baseten's $1.5B Series F at a $13B valuation on June 25, 2026
- Now: Baseten's $1.5B Series F at a $13B valuation on June 22, 2026
- Why it happened: claim audit
- Source: https://www.baseten.co/blog/announcing-our-series-f/
- Page: /answers/baseten-vs-together-ai-vs-fireworks-ai-inference-platforms-june-2026/
November 4, 2026 — Correction (1 page)
Corrected /answers/baseten-ai-infrastructure-company-evaluation-2026/
- Was: The Basic plan is $0 a month, pay as you go; Pro and Enterprise are quoted.
- Now: Self-serve is pay as you go with no contract; production capacity, volume discounts and custom terms are negotiated with sales (Baseten's pricing page no longer lists Basic/Pro/Enterprise plans).
- Why it happened: claim audit
- Source: https://www.baseten.co/pricing/
- Page: /answers/baseten-ai-infrastructure-company-evaluation-2026/
November 4, 2026 — Correction (1 page)
Corrected /answers/baseten-ai-infrastructure-company-evaluation-2026/
- Was: Reliability claim: "Four nines" for Model APIs via active-active multi-cluster autoscaling
- Now: Reliability claim: "99.9% uptime" for Model APIs, on a compute platform spanning 30+ clouds
- Why it happened: claim audit
- Source: https://www.baseten.co/products/model-apis/
- Page: /answers/baseten-ai-infrastructure-company-evaluation-2026/
November 3, 2026 — Update (1 page)
Updated /answers/best-ai-agent-control-planes-may-2026/
- Was: ServiceNow AI Control Tower
- Now: Renamed ServiceNow Super Intelligence Control Tower (October 2026), with AI Passport agent identities
- Source: https://www.servicenow.com/products/super-intelligence-control-tower.html
- Page: /answers/best-ai-agent-control-planes-may-2026/
November 3, 2026 — Update (1 page)
Updated /answers/kilo-code-vs-cline-vs-continue-open-source-coding-agents-july-2026/
- Was: Kilo Code latest release v7.8.3 (Oct 1, 2026); VS Code, JetBrains, CLI, cloud agent
- Now: Kilo Code v7.8.18 (Oct 26, 2026); Anaconda added agent swarms, native JetBrains extension, mobile apps and Sign in with ChatGPT on Oct 6, 2026
- Source: https://www.anaconda.com/products/launch/2026-october
- Page: /answers/kilo-code-vs-cline-vs-continue-open-source-coding-agents-july-2026/
November 3, 2026 — Update (1 page)
Updated /answers/ios-27-siri-extensions-hidden-claude-chatgpt-gemini-june-2026/
- Was: A later point release (27.1 or 27.2) is the earliest plausible window for Claude/Gemini Siri Extensions
- Now: iOS 27.2 shipped Oct 26, 2026 (Siri AI in 5 more languages, memory) without third-party Siri replacement; ChatGPT still the only Siri extension
- Source: https://support.apple.com/en-us/149076
- Page: /answers/ios-27-siri-extensions-hidden-claude-chatgpt-gemini-june-2026/
November 3, 2026 — Update (1 page)
Updated /answers/baseten-ai-infrastructure-company-evaluation-2026/
- Was: Funding: $1.5B Series F at $13B (June 2026), latest
- Now: Plus a reported ~$2B raise at $26B-$30B (Bloomberg, Oct 30, 2026), not yet announced by Baseten
- Source: https://www.bloomberg.com/news/articles/2026-10-30/baseten-set-for-30-billion-valuation-with-new-funding
- Page: /answers/baseten-ai-infrastructure-company-evaluation-2026/
November 3, 2026 — Update (1 page)
Updated /answers/baseten-vs-together-ai-vs-fireworks-ai-inference-platforms-june-2026/
- Was: Baseten reported to be in talks to raise at about $26B
- Now: Bloomberg (Oct 30, 2026) reports Baseten raising about $2B in two tranches at $26B and $30B; confirmed valuation still $13B
- Source: https://www.bloomberg.com/news/articles/2026-10-30/baseten-set-for-30-billion-valuation-with-new-funding
- Page: /answers/baseten-vs-together-ai-vs-fireworks-ai-inference-platforms-june-2026/
November 3, 2026 — Update (1 page)
Updated /answers/baseten-vs-together-ai-vs-fireworks-ai-inference-platforms-june-2026/
- Was: Together AI lists Kimi K3 at $2.70 / $13.50 per 1M tokens; Llama 3.3 70B at $1.04 / $1.04
- Now: Together AI lists Kimi K3 at $3.00 / $15.00 (same as Fireworks); Llama 3.3 70B is no longer on Together's serverless list
- Source: https://www.together.ai/pricing
- Page: /answers/baseten-vs-together-ai-vs-fireworks-ai-inference-platforms-june-2026/
November 3, 2026 — Update (1 page)
Added Kimiko, a local-first meeting note taker for Apple-silicon Macs and Windows 11, from the maker's public pages (features, pricing, data handling).
- Was: Six tools compared (Anarlog, Foresight, Granola, Kasset, Fireflies, Otter)
- Now: Kimiko added: local-first Mac/Windows note taker, entry, table row and decision-guide line
- Source: https://kimiko.app/your-data/
- Page: /answers/best-local-ai-meeting-note-takers-2026-foresight-anarlog-granola/
November 1, 2026 — Correction (1 page)
Corrected /answers/best-ocr-document-extraction-apis-2026-ranked/
- Was: Google Document AI: the basic Read API is $0.65 (per 1,000 pages)
- Now: Removed: Google's Document AI pricing page lists no $0.65 Read API; its cheapest OCR is Enterprise Document OCR at $1.50 per 1,000 pages ($0.60 above 5 million)
- Why it happened: claim audit
- Source: https://cloud.google.com/document-ai/pricing
- Page: /answers/best-ocr-document-extraction-apis-2026-ranked/
November 1, 2026 — Correction (1 page)
Corrected /answers/best-ocr-document-extraction-apis-2026-ranked/
- Was: Mistral OCR 4 throughput is up to ~2,000 pages per minute per GPU
- Now: Removed: Mistral's OCR 4 announcement gives no throughput figure (the 2,000 pages a minute figure was for the earlier Mistral OCR, per node)
- Why it happened: claim audit
- Source: https://mistral.ai/news/ocr-4/
- Page: /answers/best-ocr-document-extraction-apis-2026-ranked/
November 1, 2026 — Correction (1 page)
Corrected /answers/best-ocr-document-extraction-apis-2026-ranked/
- Was: Azure Document Intelligence custom classification $50 per 1,000 pages
- Now: Azure Document Intelligence custom classification $3 per 1,000 pages
- Why it happened: claim audit
- Source: https://prices.azure.com/api/retail/prices?$filter=armRegionName%20eq%20'eastus'%20and%20productName%20eq%20'Azure%20Document%20Intelligence'
- Page: /answers/best-ocr-document-extraction-apis-2026-ranked/
October 31, 2026 — Update (1 page)
Added Extend's data-retention terms from its pricing page
- Was: Extend entry did not state its data-retention terms
- Now: Extend entry notes zero data retention on pay-as-you-go (available on request per the plan comparison; included on Growth and Enterprise)
- Source: https://www.extend.ai/pricing
- Page: /answers/best-ocr-document-extraction-apis-2026-ranked/
October 30, 2026 — Update (1 page)
Updated /answers/enterprise-work-platforms-ai-governance-data-privacy-comparison-2026/
- Was: Five enterprise platforms compared on AI governance and data privacy.
- Now: Added a section on GoodDay: its AI features, MCP server, published security and GDPR controls, plan pricing, and the AI data-handling questions its public pages do not answer.
- Source: https://www.goodday.work/security
- Page: /answers/enterprise-work-platforms-ai-governance-data-privacy-comparison-2026/
October 30, 2026 — Update (1 page)
Updated /answers/best-ocr-document-extraction-apis-2026-ranked/
- Was: Seven ranked OCR and document extraction APIs.
- Now: Added a section on Extend (parse, extract, split and classify API): credit pricing per 1,000 pages, free credits and its published RealDoc-Bench results, noted as the vendor's own benchmark.
- Source: https://www.extend.ai/pricing
- Page: /answers/best-ocr-document-extraction-apis-2026-ranked/
October 30, 2026 — Correction (1 page)
Corrected /posts/omniroute-free-ai-gateway-llm-providers-review/
- Was: The compression alone saves ~$40/month in tokens.
- Now: Removed the unsourced $40/month savings figure: the compression pays off most on tool-heavy sessions (build logs, diffs).
- Why it happened: claim audit
- Source: https://github.com/cheaper-inference/OmniRoute
- Page: /posts/omniroute-free-ai-gateway-llm-providers-review/
October 30, 2026 — Correction (1 page)
Corrected /posts/claude-video-watch-skill-frames-transcribe-review/
- Was: Groq's whisper-large-v3 is roughly half a cent per five-minute clip
- Now: Groq's whisper-large-v3 lists at $0.111 per audio hour, roughly one cent per five-minute clip
- Why it happened: claim audit
- Source: https://console.groq.com/docs/speech-to-text
- Page: /posts/claude-video-watch-skill-frames-transcribe-review/
October 30, 2026 — Correction (1 page)
Corrected /answers/inference-providers-sub-second-cold-start-2026/
- Was: Runpod Serverless H100 flex worker $4.79/h; per-second rates up to $10.65/h for a B300
- Now: Runpod Serverless H100 flex worker $5.39/h; per-second rates up to $11.99/h for a B300
- Why it happened: claim audit
- Source: https://www.runpod.io/pricing
- Page: /answers/inference-providers-sub-second-cold-start-2026/
October 29, 2026 — Update (1 page)
Updated /posts/taste-skill-anti-slop-ai-frontend-review/
- Was: v2 still labelled experimental; 13-skill bundle unchanged
- Now: v2.0.0 stable shipped Oct 13, 2026 (latest v2.1.5, Oct 24); repo now holds one skill, the others moved to taste-skill-classics and imagegen-skills
- Source: https://github.com/Leonxlnx/taste-skill/releases
- Page: /posts/taste-skill-anti-slop-ai-frontend-review/
October 29, 2026 — Update (1 page)
Updated /posts/anythingllm-all-in-one-ai-app/
- Was: Latest release v1.16.1 (Aug 27, 2026)
- Now: Latest release v1.19.0 (Oct 26, 2026), adding an opt-in built-in LLM for Docker (ENABLE_BUILT_IN_LLM=true)
- Source: https://github.com/Mintplex-Labs/anything-llm/releases
- Page: /posts/anythingllm-all-in-one-ai-app/
October 29, 2026 — Update (1 page)
Updated /posts/claude-video-watch-skill-frames-transcribe-review/
- Was: v0.2.0 is still the latest tag; code unchanged since July 1, 2026; Whisper transcription API-only; watch.skill uploadable to claude.ai
- Now: v0.3.0-v0.3.2 shipped Sep 25, 2026: Gemini engine, optional local WhisperX transcription; Claude Chat and Cowork no longer supported
- Source: https://github.com/bradautomates/claude-video/blob/main/CHANGELOG.md
- Page: /posts/claude-video-watch-skill-frames-transcribe-review/
October 29, 2026 — Correction (1 page)
Corrected /posts/omniroute-free-ai-gateway-llm-providers-review/
- Was: Review said the author ran OmniRoute v3.8.44 for four days and presented compression figures as 'my four-day test results'
- Now: No recorded hands-on run exists; the review now says it was written from the README and docs, and the compression table is labelled illustrative, not measured by us
- Why it happened: Hands-on claim without an evidence record (EVIDENCE-SPEC.md)
- Source: https://github.com/cheaper-inference/OmniRoute
- Page: /posts/omniroute-free-ai-gateway-llm-providers-review/
October 29, 2026 — Update (1 page)
Updated /posts/omniroute-free-ai-gateway-llm-providers-review/
- Was: Kiro listed as a free unlimited emergency fallback; Claude Code setup via omniroute setup-claude-code
- Now: Kiro is marked tos: avoid and excluded from auto routing by default; Claude Code setup is omniroute setup-claude, or omniroute launch
- Source: https://github.com/cheaper-inference/OmniRoute/blob/main/docs/guides/CLI-INTEGRATIONS.md
- Page: /posts/omniroute-free-ai-gateway-llm-providers-review/
October 29, 2026 — Update (1 page)
Updated /posts/omniroute-free-ai-gateway-llm-providers-review/
- Was: Repository at github.com/diegosouzapw/OmniRoute; 237 providers, 17 routing strategies, 95 MCP tools
- Now: Repository moved to github.com/cheaper-inference/OmniRoute; README claims 377 providers, 20 routing strategies, 110 MCP tools; latest release v3.8.52 (Oct 19, 2026)
- Source: https://github.com/cheaper-inference/OmniRoute
- Page: /posts/omniroute-free-ai-gateway-llm-providers-review/
October 29, 2026 — Correction (1 page)
Corrected /answers/what-is-entire-git-forge-dohmke-trails-2026/
- Was: No pricing page at launch; launch coverage reported a free start, a $100 credit for Trails review agents, and tokens billed at market rates
- Now: Entire's billing docs: only Agent usage is billed (prepaid credits), other features free; one-time $100 trial credit per billing account, then Agent tokens at Entire's listed per-model rates
- Why it happened: claim audit
- Source: https://docs.entire.com/billing
- Page: /answers/what-is-entire-git-forge-dohmke-trails-2026/
October 28, 2026 — Correction (1 page)
Corrected /posts/genoffice-open-source-ai-office-suite-review/
- Was: 73 open issues
- Now: 56 open issues (excluding pull requests)
- Why it happened: claim audit
- Source: https://github.com/genspark-ai/genoffice
- Page: /posts/genoffice-open-source-ai-office-suite-review/
October 28, 2026 — Correction (1 page)
Corrected /answers/what-is-entire-git-forge-dohmke-trails-2026/
- Was: regions in the US (Ohio, Oregon), EU (Frankfurt, Ireland), India (Mumbai), Australia
- Now: primary region US (Ohio), mirrored regions EU (Frankfurt), India (Mumbai), Australia (Sydney)
- Why it happened: claim audit
- Source: https://entire.com/
- Page: /answers/what-is-entire-git-forge-dohmke-trails-2026/
October 28, 2026 — Update (1 page)
Updated /posts/genoffice-open-source-ai-office-suite-review/
- Was: Six Electron apps: Docs, Sheets, Slides, PDF, Markdown and a tabbed shell (true at v0.8.0)
- Now: Update note: an HTML app joined the suite, making seven Electron apps
- Why it happened: claim audit
- Source: https://github.com/genspark-ai/genoffice
- Page: /posts/genoffice-open-source-ai-office-suite-review/
October 28, 2026 — Correction (1 page)
Corrected /answers/github-copilot-agent-mode-vs-claude-code-vs-openai-codex-2026/
- Was: Every paid Copilot plan can delegate tasks to Claude and OpenAI Codex as third-party coding agents
- Now: Copilot Pro+ and Max can delegate to Claude and Codex as third-party agents; the $10 Pro plan cannot
- Why it happened: claim audit
- Source: https://github.com/features/copilot/plans
- Page: /answers/github-copilot-agent-mode-vs-claude-code-vs-openai-codex-2026/
October 28, 2026 — Correction (1 page)
Corrected /answers/qwen-image-2-1-vs-flux-2-dev-vs-gpt-image-2-5-vs-nano-banana-pro-september-2026/
- Was: Alibaba Cloud pricing TBA
- Now: Pro and Turbo APIs live on Alibaba Cloud Model Studio since October 9, 2026
- Why it happened: claim audit
- Source: https://github.com/QwenLM/Qwen-Image-2.1
- Page: /answers/qwen-image-2-1-vs-flux-2-dev-vs-gpt-image-2-5-vs-nano-banana-pro-september-2026/
October 28, 2026 — Correction (1 page)
Corrected /answers/what-is-gpt-6-1-terra-openai-price-computer-use-2026/
- Was: Artificial Analysis lists GPT-6.1 Terra at 48 on its Intelligence Index, level with GPT-6 Sol, below GPT-6.1 Sol (52), GPT-6 Astra (53) and Claude Opus 5.5 (58)
- Now: Artificial Analysis lists GPT-6.1 Terra (max) at 51, below GPT-6.1 Sol (55) and GPT-6 Astra (57)
- Why it happened: claim audit
- Source: https://artificialanalysis.ai/models/releases/gpt-6-1-terra
- Page: /answers/what-is-gpt-6-1-terra-openai-price-computer-use-2026/
October 28, 2026 — Correction (1 page)
Corrected /answers/enterprise-work-platforms-ai-governance-data-privacy-comparison-2026/
- Was: Salesforce's Einstein Trust Layer masks sensitive fields before a prompt is sent (incl. Agentforce)
- Now: Masking applies to embedded generative AI features such as Einstein Service Replies; Salesforce says data masking is currently disabled for agents
- Why it happened: claim audit
- Source: https://trailhead.salesforce.com/content/learn/modules/the-einstein-trust-layer/meet-the-einstein-trust-layer
- Page: /answers/enterprise-work-platforms-ai-governance-data-privacy-comparison-2026/
October 27, 2026 — Update (1 page)
Updated /answers/best-ai-agent-payment-solutions-2026/
- Was: x402 is Coinbase's protocol; Stripe MPP launched days ago; five options (x402, MPP, Crossmint, Visa, World AgentKit)
- Now: x402 governed by the x402 Foundation (Linux Foundation, operational Jul 14, 2026); added Mastercard Agent Pay for Machines (early access Oct 2026), Google AP2 and Stripe ACP; Crossmint Agent Commerce Toolkit (Oct 2026)
- Source: https://x402.org/linux-foundation-announces-operational-launch-of-x402-foundation-to-standardize-internet-native-payments-for-ai-agents-and-applications/
- Page: /answers/best-ai-agent-payment-solutions-2026/
October 27, 2026 — Update (1 page)
Updated /posts/genoffice-open-source-ai-office-suite-review/
- Was: Latest v0.8.0 (Aug 23, 2026); 3,566 stars; '[Docs] Layout breaks when opening .docx' issue open; no Windows on Arm build
- Now: Latest v0.12.115 (Oct 26, 2026); 11,052 stars; layout issue closed; native Windows on Arm installer; full RTL support still open (issue #13)
- Source: https://github.com/genspark-ai/genoffice/releases
- Page: /posts/genoffice-open-source-ai-office-suite-review/
October 27, 2026 — Update (1 page)
Updated /posts/orca-stablyai-parallel-coding-agents-ide-review/
- Was: Linux is AppImage only; latest v1.4.196 (Sep 3, 2026); 60,400+ stars
- Now: Linux ships AppImage, .deb, .rpm and ARM64 builds; latest v1.4.242 (Oct 27, 2026); 107,000+ stars
- Source: https://github.com/stablyai/orca/releases
- Page: /posts/orca-stablyai-parallel-coding-agents-ide-review/
October 27, 2026 — Correction (1 page)
Corrected /posts/orca-stablyai-parallel-coding-agents-ide-review/
- Was: Comparison table and FAQ: Orca is open source under Apache 2.0
- Now: Orca is MIT-licensed (table and FAQ now match the September correction)
- Source: https://github.com/stablyai/orca
- Page: /posts/orca-stablyai-parallel-coding-agents-ide-review/
October 27, 2026 — Correction (1 page)
Corrected /answers/qwen-image-2-1-vs-flux-2-dev-vs-gpt-image-2-5-vs-nano-banana-pro-september-2026/
- Was: Nano Banana Pro API price ~$0.039 (1K) – $0.24 (4K) per image
- Now: Nano Banana Pro $0.134 per 1K/2K image and $0.24 per 4K image; Nano Banana 2.1 ~$0.034 (1K) – $0.113 (4K)
- Source: https://ai.google.dev/gemini-api/docs/pricing
- Page: /answers/qwen-image-2-1-vs-flux-2-dev-vs-gpt-image-2-5-vs-nano-banana-pro-september-2026/
October 27, 2026 — Update (1 page)
Updated /answers/github-copilot-agent-mode-vs-claude-code-vs-openai-codex-2026/
- Was: Codex CLI API pricing $5/$30 per MTok (GPT-5.6 Sol) to $0.15/$0.60 (GPT-5 Mini)
- Now: GPT-6 Astra $10/$50, GPT-6.1 Sol $2/$10, GPT-6.1 Terra $1/$5, GPT-6 Luna $0.10/$0.50; GPT-5.6 Sol $4/$20 (promo); GPT-5 mini is $0.25/$2.00
- Source: https://developers.openai.com/api/docs/pricing
- Page: /answers/github-copilot-agent-mode-vs-claude-code-vs-openai-codex-2026/
October 27, 2026 — Update (1 page)
Updated /answers/github-copilot-agent-mode-vs-claude-code-vs-openai-codex-2026/
- Was: Claude Code runs Claude Sonnet 5 and Opus 4.8 with a 200K context window; Copilot runs GPT-4.1 and fine-tuned models; Codex runs GPT-5.6 Sol/Terra/Luna and GPT-5.5
- Now: Claude Code runs Opus 5.5 / Sonnet 5.5 (1M context); Copilot offers Claude, GPT-6.x, Gemini 4 Argon, Grok 4.7, Kimi K3 and can delegate to Claude and Codex; Codex runs GPT-6 Astra, GPT-6.1 Sol/Terra, GPT-6 Luna (GPT-5.5 retired from Codex Oct 14, 2026)
- Source: https://github.com/features/copilot/plans
- Page: /answers/github-copilot-agent-mode-vs-claude-code-vs-openai-codex-2026/
October 27, 2026 — Correction (1 page)
Corrected /answers/github-copilot-agent-mode-vs-claude-code-vs-openai-codex-2026/
- Was: Claude Code API pricing: $0.003 per input token / $0.015 per output token for Sonnet 5
- Now: Per million tokens: Opus 5.5 $4/$20, Sonnet 5.5 $2/$10, Haiku 5.5 $0.10/$0.50; Claude Code included in Pro ($20/mo) and Max ($100/$200/mo)
- Source: https://claude.com/pricing
- Page: /answers/github-copilot-agent-mode-vs-claude-code-vs-openai-codex-2026/
October 27, 2026 — Correction (1 page)
Corrected /answers/github-copilot-agent-mode-vs-claude-code-vs-openai-codex-2026/
- Was: Codex CLI was released in early 2026
- Now: Codex CLI was first released in April 2025
- Source: https://github.com/openai/codex
- Page: /answers/github-copilot-agent-mode-vs-claude-code-vs-openai-codex-2026/
October 27, 2026 — Correction (1 page)
Corrected /answers/github-copilot-agent-mode-vs-claude-code-vs-openai-codex-2026/
- Was: GitHub Copilot Agent Mode launched in February 2026
- Now: Copilot agent mode was introduced in VS Code (preview) in February 2025 and rolled out to all VS Code users in April 2025
- Source: https://github.blog/news-insights/product-news/github-copilot-the-agent-awakens/
- Page: /answers/github-copilot-agent-mode-vs-claude-code-vs-openai-codex-2026/
October 27, 2026 — Update (1 page)
Added type.com to the agent platforms comparison.
- Was: Six platforms compared: n8n, Zapier, Copilot Studio, LangGraph + LangSmith, CrewAI, OpenClaw.
- Now: type.com added: a table row and a situation entry (managed runtime for team Claude/Codex agent automations; plans, credits, limits from type.com's pricing page and docs).
- Source: https://type.com/pricing
- Page: /answers/best-ai-agent-platforms-for-autonomous-workflows-2026/
October 26, 2026 — Update (1 page)
Added the pre-push preview command (uiverify check) to the hosted visual-diff option, from its documentation.
- Was: The UI Verify note covered pull-request builds only.
- Now: The note also covers uiverify check, which previews an edit against the CI baseline before a pull request.
- Source: https://uiverify.ai/docs/check-visual-changes
- Page: /answers/how-to-verify-ai-agent-ui-output/
October 26, 2026 — Update (1 page)
Added a bot-free cloud note taker that stores data in the EU (Kasset) to the comparison, with prices and data handling from the maker's own pages.
- Was: Five note takers compared: Anarlog, Google AI Edge Foresight, Granola, Fireflies and Otter.
- Now: Kasset added as a bot-free cloud option with data stored in France, with prices and data handling from the maker's own pages.
- Source: https://kasset.com/gdpr-ai-note-taker/
- Page: /answers/best-local-ai-meeting-note-takers-2026-foresight-anarlog-granola/
October 26, 2026 — Update (1 page)
Added a hosted visual-diff option (UI Verify) to the baseline section, with pricing from the maker's site and a check of its public demo.
- Was: Solution 3 (visual diff against a baseline) covered only a local ProofShot workflow; no Sources section.
- Now: Added UI Verify as a hosted baseline-diff option (capture methods, AI judge, MCP server, October 2026 pricing), an FAQ entry and a Sources section.
- Source: https://uiverify.ai/pricing
- Page: /answers/how-to-verify-ai-agent-ui-output/
October 26, 2026 — Update (1 page)
Updated /answers/claude-code-weekly-limits-extended-august-31-2026/
- Was: Gemini 3.7 Flash at $0.75/$3.75 or GLM-5.3 at $1.40/$4.40 per MTok
- Now: Gemini 3.8 Flash at $0.75/$3.75 per MTok (introductory through December 31, 2026); Gemini 3.7 Flash was deprecated on October 8, 2026 and now routes to 3.8 Flash
- Why it happened: claim audit
- Source: https://ai.google.dev/gemini-api/docs/pricing
- Page: /answers/claude-code-weekly-limits-extended-august-31-2026/
October 26, 2026 — Correction (1 page)
Corrected /answers/claude-code-weekly-limits-extended-august-31-2026/
- Was: Fable 5 access is restricted to Max and premium Team and seat-based Enterprise seats
- Now: Fable is capped at 50% of the weekly allowance on Max and premium Team seats, included on Enterprise plans, and available on Pro only through paid usage credits
- Why it happened: claim audit
- Source: https://www.anthropic.com/pricing
- Page: /answers/claude-code-weekly-limits-extended-august-31-2026/
October 25, 2026 — Correction (1 page)
Updated the end date of the Claude Code weekly limit boost and replaced a dead Anthropic source link.
- Was: The page gave August 31, 2026 as the current end date of the +50% Claude Code weekly limit boost and cited an Anthropic help article that now returns 404.
- Now: The page notes that the boost was extended through September 13, 2026 and replaced on September 14, 2026 by a permanent 25% increase over the pre-promotion limits; the help-article link points to Anthropic's current usage-limits article.
- Why it happened: A dated news page was not revisited when the promotion changed again; reported by a reader.
- Source: https://www.bleepingcomputer.com/news/artificial-intelligence/anthropic-is-cutting-claude-codes-current-weekly-limits-by-17-percent/
- Page: /answers/claude-code-weekly-limits-extended-august-31-2026/
October 21, 2026 — Update (1 page)
Scribbl connector and remote-connector setup added; FAQ and Sources added
- Was: Listed local MCP servers by category; no FAQ or Sources section
- Now: Added Scribbl's remote meeting connector (read-only search and read of recorded meetings, included with paid Scribbl plans) to the productivity table and task table, a note on adding remote MCP servers as custom connectors, three FAQ entries and a Sources section
- Source: https://scribbl.co/ai-connector
- Page: /answers/best-mcp-tools-for-claude-2026/
October 21, 2026 — Update (1 page)
type.com added to the alternatives; three type.com sources added
- Was: Compared AgentCore with LangSmith Deployment, Claude Managed Agents, Google's agent platform, Foundry and self-hosting
- Now: Added type.com, a hosted runtime for team agent automations (managed Claude or Codex agents in sandboxed VMs; $50/$100/$250 a month by team size, AI credits included), to the table, the 'when each one wins' section, the pick and the FAQ; sources appended
- Source: https://type.com/pricing
- Page: /answers/aws-bedrock-agentcore-alternatives-multi-cloud-2026/
October 21, 2026 — Correction (1 page)
Corrected /posts/herdr-agent-multiplexer-terminal-review/
- Was: I installed Herdr via Homebrew and tested it across a day of real work; memory under 15MB for 6 panes and 3 agents; new session under 100ms; setup under 10 seconds
- Now: Not run by us: the section describes what the README and docs state; memory, CPU and startup time were not measured
- Why it happened: claim audit
- Source: https://github.com/herdrdev/herdr
- Page: /posts/herdr-agent-multiplexer-terminal-review/
October 21, 2026 — Correction (1 page)
Corrected /answers/cloud-providers-real-time-audio-streaming-llm-apps-2026/
- Was: LiveKit's calculator puts STT, LLM and TTS at roughly $0.02 to $0.04 a minute
- Now: LiveKit's calculator puts its sample cascaded stack at about $0.018 a minute (LLM $0.0014, STT $0.0075, TTS $0.0090)
- Why it happened: claim audit
- Source: https://livekit.com/pricing
- Page: /answers/cloud-providers-real-time-audio-streaming-llm-apps-2026/
October 21, 2026 — Correction (1 page)
Corrected /answers/backfile-conversion-pilot-project-sample-pricing-us-2026/
- Was: a two-box pilot of about 4,400 pages runs roughly $400–$700 in scanning and OCR
- Now: a two-box pilot of about 4,400 pages runs roughly $400–$570 in scanning and OCR (4,400 pages at 9–13 cents)
- Why it happened: claim audit
- Source: https://www.accesscorp.com/blog/how-much-does-backfile-conversion-cost-and-why/
- Page: /answers/backfile-conversion-pilot-project-sample-pricing-us-2026/
October 21, 2026 — Correction (1 page)
Corrected /answers/how-to-secure-ai-agent-credentials-2026-guide/
- Was: An agent entered through a malicious dataset exploiting a remote-code dataset loader and a template-injection flaw
- Now: Agents used 14 publicly exposed Hugging Face credentials and a malicious dataset exploiting two code-execution flaws (Hugging Face: remote-code dataset loader and template injection; OpenAI: HDF5 file-handling zero-day and RefJinja template-injection zero-day)
- Why it happened: claim audit
- Source: https://openai.com/index/hugging-face-incident-and-the-road-ahead/
- Page: /answers/how-to-secure-ai-agent-credentials-2026-guide/
October 21, 2026 — Correction (1 page)
Corrected /answers/best-api-providers-muse-spark-open-nemotron-3-5-super-vl-2026/
- Was: Baseten runs Nemotron 3.5 Super VL as a launch partner (no price listed); CoreWeave is the only listed host with the full ~1M-token context
- Now: Baseten lists Nemotron 3.5 Super VL at $0.20 input / $0.04 cached / $0.50 output per million tokens with a 1,048,576-token context (FP4); CoreWeave and Baseten both serve the full ~1M window
- Why it happened: claim audit
- Source: https://openrouter.ai/nvidia/nemotron-3.5-super-vl
- Page: /answers/best-api-providers-muse-spark-open-nemotron-3-5-super-vl-2026/
October 20, 2026 — Correction (1 page)
Corrected /answers/what-is-unconventional-ai-un-0-oscillator-model-june-2026/
- Was: Seed round led by Andreessen Horowitz
- Now: Seed round co-led by Andreessen Horowitz and Lightspeed Venture Partners
- Source: https://techcrunch.com/2025/12/09/unconventional-ai-confirms-its-massive-475m-seed-round/
- Page: /answers/what-is-unconventional-ai-un-0-oscillator-model-june-2026/
October 20, 2026 — Update (1 page)
Updated /answers/what-is-unconventional-ai-un-0-oscillator-model-june-2026/
- Was: No physical chip yet; system-on-chip tape-in planned later in 2026
- Now: First chip taped out June 1, 2026 and back from the fab with working results (CEO, All-In Summit, September 2026); rack-scale product targeted within ~2 years
- Source: https://finance.yahoo.com/video/summit-unconventional-ai-ceo-naveen-194251805.html
- Page: /answers/what-is-unconventional-ai-un-0-oscillator-model-june-2026/
October 20, 2026 — Update (1 page)
Updated /posts/herdr-agent-multiplexer-terminal-review/
- Was: Current version v0.4.0; repo ogulcancelik/herdr
- Now: Current version v0.9.5 (October 14, 2026); repo moved to herdrdev/herdr
- Source: https://github.com/herdrdev/herdr/releases/tag/v0.9.5
- Page: /posts/herdr-agent-multiplexer-terminal-review/
October 20, 2026 — Update (1 page)
Updated /posts/herdr-agent-multiplexer-terminal-review/
- Was: Herdr is AGPL-3.0 (commercial licenses available)
- Now: Herdr is Apache-2.0 since July 22, 2026
- Source: https://github.com/herdrdev/herdr/blob/main/LICENSE
- Page: /posts/herdr-agent-multiplexer-terminal-review/
October 20, 2026 — Update (1 page)
Updated /answers/apple-core-ai-vs-foundation-models-vs-mlx-ios-27-framework-june-2026/
- Was: iOS 27 is in developer beta; APIs may change before fall release
- Now: iOS 27 publicly released September 14, 2026
- Source: https://www.apple.com/newsroom/2026/09/major-updates-for-apples-software-platforms-are-now-available/
- Page: /answers/apple-core-ai-vs-foundation-models-vs-mlx-ios-27-framework-june-2026/
October 20, 2026 — Correction (1 page)
Corrected /answers/apple-core-ai-vs-foundation-models-vs-mlx-ios-27-framework-june-2026/
- Was: MLX has direct export paths to Core AI
- Now: Apple documents no MLX-to-Core AI export; the supported path is PyTorch -> coreai-torch -> .aimodel
- Source: https://developer.apple.com/core-ai/
- Page: /answers/apple-core-ai-vs-foundation-models-vs-mlx-ios-27-framework-june-2026/
October 20, 2026 — Correction (1 page)
Corrected /answers/apple-core-ai-vs-foundation-models-vs-mlx-ios-27-framework-june-2026/
- Was: Core AI model files use the .aiasset format
- Now: Core AI model files are .aimodel (compiled: .aimodelc); AIModelAsset is the Swift type
- Source: https://developer.apple.com/documentation/coreai/integrating-on-device-ai-models-in-your-app-with-core-ai
- Page: /answers/apple-core-ai-vs-foundation-models-vs-mlx-ios-27-framework-june-2026/
October 20, 2026 — Update (1 page)
Updated /posts/apple-container-mac-linux-docker-alternative-review/
- Was: 1.4.1 is the current apple/container release (September 2026)
- Now: 1.6.0 is current (October 20, 2026); 1.5.0 added a k8s-plugin security fix and removed 'container k8s start'
- Source: https://github.com/apple/container/releases/tag/1.6.0
- Page: /posts/apple-container-mac-linux-docker-alternative-review/
October 20, 2026 — Correction (1 page)
Corrected /answers/best-api-providers-muse-spark-open-nemotron-3-5-super-vl-2026/
- Was: CoreWeave showed 96.4% recent uptime on OpenRouter, against 100% for the others
- Now: OpenRouter's one-day uptime on October 20, 2026: Crusoe under 90%, CoreWeave about 98%, Together, DeepInfra and Venice 100%
- Why it happened: claim audit
- Source: https://openrouter.ai/nvidia/nemotron-3.5-super-vl
- Page: /answers/best-api-providers-muse-spark-open-nemotron-3-5-super-vl-2026/
October 20, 2026 — Correction (1 page)
Corrected /answers/best-api-providers-muse-spark-open-nemotron-3-5-super-vl-2026/
- Was: DeepInfra (FP4) listed for Nemotron 3.5 Super VL at $0.10 input, $0.08 cached and $0.40 output per million tokens, tied with Crusoe as cheapest
- Now: DeepInfra (FP4) lists $0.15 input, $0.02 cached and $0.50 output per million tokens; Crusoe alone is cheapest at $0.10/$0.40
- Why it happened: claim audit
- Source: https://openrouter.ai/nvidia/nemotron-3.5-super-vl
- Page: /answers/best-api-providers-muse-spark-open-nemotron-3-5-super-vl-2026/
October 19, 2026 — Correction (1 page)
Corrected /answers/servicenow-ai-control-tower-alternatives-2026/
- Was: IBM watsonx.governance: Essentials and Standard tiers; quote-based
- Now: IBM watsonx.governance: free trial; Model Management from $0.64 pay as you go; Risk & Compliance from $3,500 a month
- Why it happened: claim audit
- Source: https://www.ibm.com/products/watsonx-governance/pricing
- Page: /answers/servicenow-ai-control-tower-alternatives-2026/
October 19, 2026 — Correction (1 page)
Corrected /answers/servicenow-ai-control-tower-alternatives-2026/
- Was: since October 15, 2026, issues AI Passports
- Now: from later in the fourth quarter of 2026 (announced October 15, 2026), will issue AI Passports
- Why it happened: claim audit
- Source: https://siliconangle.com/2026/10/15/servicenow-adds-ai-agent-controls-and-employee-automation-tools/
- Page: /answers/servicenow-ai-control-tower-alternatives-2026/
October 18, 2026 — Correction (1 page)
Corrected /answers/best-inference-provider-gpt-oss-120b-production-2026/
- Was: Cerebras roughly 3.5x Groq and nearly 20x the slowest GPU hosts
- Now: Cerebras roughly 3.5x Groq and about 41x the slowest host Artificial Analysis measures (DeepInfra, 42 tok/s)
- Why it happened: claim audit
- Source: https://artificialanalysis.ai/models/gpt-oss-120b/providers
- Page: /answers/best-inference-provider-gpt-oss-120b-production-2026/
October 17, 2026 — Update (1 page)
Updated /answers/best-inference-provider-gpt-oss-120b-production-2026/
- Was: Speeds: Cerebras ~1,708 tok/s, 0.43 s; Groq 0.73 s; Together ~89 tok/s, 0.56 s; Bedrock ~88 tok/s; Baseten ~253 tok/s, 0.20 s
- Now: Re-read Oct 17, 2026: Cerebras ~1,738 tok/s, 0.42 s; Groq 0.70 s; Together ~74 tok/s, 0.58 s; Bedrock ~87 tok/s; Baseten ~238 tok/s, 0.19 s
- Why it happened: claim audit
- Source: https://artificialanalysis.ai/models/gpt-oss-120b/providers
- Page: /answers/best-inference-provider-gpt-oss-120b-production-2026/
October 17, 2026 — Correction (1 page)
Corrected /answers/best-inference-provider-gpt-oss-120b-production-2026/
- Was: more than 20 hosts serve it
- Now: Artificial Analysis benchmarks 16 API hosts serving it (Fireworks is not among them)
- Why it happened: claim audit
- Source: https://artificialanalysis.ai/models/gpt-oss-120b/providers
- Page: /answers/best-inference-provider-gpt-oss-120b-production-2026/
October 17, 2026 — Correction (1 page)
Corrected /answers/best-inference-provider-gpt-oss-120b-production-2026/
- Was: Budget hosts CoreWeave, DeepInfra and Novita listed at blended $0.02–$0.04 per million tokens; Baseten 'blended ~$0.07' (these were Artificial Analysis cost-per-task figures)
- Now: Blended $0.04–$0.07 per million tokens (CoreWeave $0.04, DeepInfra $0.05, Novita $0.07); Baseten's $0.07 relabelled as AA cost per benchmark task
- Why it happened: claim audit
- Source: https://artificialanalysis.ai/models/gpt-oss-120b/providers
- Page: /answers/best-inference-provider-gpt-oss-120b-production-2026/
October 16, 2026 — Update (1 page)
Updated /answers/claude-opus-5-5-vs-claude-fable-5-1-vs-gpt-6-astra-september-2026/
- Was: Cyber Verification Program for Opus 5.5 opening in the coming weeks
- Now: Updated Cyber Verification Program open via Anthropic's Verification Portal, covering Opus 5.5, Sonnet 5.5 and Mythos 5.1
- Source: https://support.claude.com/en/articles/14604842
- Page: /answers/claude-opus-5-5-vs-claude-fable-5-1-vs-gpt-6-astra-september-2026/
October 16, 2026 — Update (1 page)
Updated /answers/rtx-spark-vs-apple-m5-max-vs-amd-ryzen-ai-max-june-2026/
- Was: RTX Spark ships fall 2026, estimated $2,500+
- Now: RTX Spark laptops on sale October 16, 2026 from $2,599 (ASUS, Dell, HP, Lenovo, Microsoft, MSI)
- Source: https://nvidianews.nvidia.com/news/nvidia-rtx-spark-pcs-available-today-a-new-beginning-for-windows-pcs
- Page: /answers/rtx-spark-vs-apple-m5-max-vs-amd-ryzen-ai-max-june-2026/
October 16, 2026 — Correction (1 page)
Corrected /answers/rtx-spark-vs-apple-m5-max-vs-amd-ryzen-ai-max-june-2026/
- Was: RTX Spark runs Llama 5 70B in FP16 and a 671B DeepSeek MoE natively in FP4
- Now: 70B FP16 (~140GB) and 671B MoE at 4-bit (~335GB) do not fit in 128GB on any of the three chips; NVIDIA rates 128GB RTX Spark for models up to 120B parameters
- Source: https://www.nvidia.com/en-us/products/rtx-spark/
- Page: /answers/rtx-spark-vs-apple-m5-max-vs-amd-ryzen-ai-max-june-2026/
October 16, 2026 — Correction (1 page)
Corrected /answers/rtx-spark-vs-apple-m5-max-vs-amd-ryzen-ai-max-june-2026/
- Was: Apple M5 Max: 16-core CPU, ~546 GB/s, from $3,499
- Now: Apple M5 Max: 18-core CPU, 32- or 40-core GPU, 460 or 614 GB/s; 14-inch MacBook Pro with M5 Max from $3,599
- Source: https://www.apple.com/newsroom/2026/03/apple-introduces-macbook-pro-with-all-new-m5-pro-and-m5-max/
- Page: /answers/rtx-spark-vs-apple-m5-max-vs-amd-ryzen-ai-max-june-2026/
October 16, 2026 — Correction (1 page)
Corrected /answers/what-is-microsoft-foundry-agent-platform-may-2026/
- Was: Azure AI Foundry was rebranded Microsoft Foundry in early 2026
- Now: Azure AI Foundry was renamed Microsoft Foundry at Microsoft Ignite in November 2025
- Source: https://azure.microsoft.com/en-us/blog/microsoft-foundry-scale-innovation-on-a-modular-interoperable-and-secure-agent-stack/
- Page: /answers/what-is-microsoft-foundry-agent-platform-may-2026/
October 16, 2026 — Update (1 page)
Updated /answers/how-to-get-claude-mythos-5-access-june-2026/
- Was: Mythos access only via Project Glasswing through an account team (4-12 weeks) or [email protected]; no plan inclusion
- Now: Mythos 5.1 is current; apply in Anthropic's Verification Portal to the Cyber Verification Program (decision targeted in 7 business days, absorbed Glasswing) or the Life Sciences Verification Program; Mythos 5.1 follows Fable 5.1 usage rules on Pro/Max for approved users
- Source: https://support.claude.com/en/articles/14604842
- Page: /answers/how-to-get-claude-mythos-5-access-june-2026/
October 16, 2026 — Update (1 page)
Updated /answers/devin-vs-claude-code-vs-cursor-april-2026/
- Was: Claude Code models Opus 4.6 $5/$25 and Sonnet 4.6 $3/$15 per M tokens
- Now: Current models Opus 5.5 $4/$20 and Sonnet 5.5 $2/$10 per M tokens
- Source: https://platform.claude.com/docs/en/about-claude/pricing
- Page: /answers/devin-vs-claude-code-vs-cursor-april-2026/
October 16, 2026 — Update (1 page)
Updated /answers/devin-vs-claude-code-vs-cursor-april-2026/
- Was: Cursor: Free, Pro $20/mo (500 fast requests), Business $40/mo/seat; multi-model included
- Now: Cursor sold as xAI's X Subscription: Free, Plus $8, Premium $30, Super $100, Ultra $200; third-party models (Claude, GPT) billed separately at API list price
- Source: https://cursor.com/pricing
- Page: /answers/devin-vs-claude-code-vs-cursor-april-2026/
October 16, 2026 — Update (1 page)
Updated /answers/devin-vs-claude-code-vs-cursor-april-2026/
- Was: Devin: Core $20/mo + ACUs, Team $500/mo, extra ACUs ~$2.25
- Now: Devin self-serve: Free, Pro $20/mo, Max $200/mo, Teams from $80/mo ($40 per full seat); ACU billing only on Enterprise
- Source: https://docs.devin.ai/admin/billing/self-serve
- Page: /answers/devin-vs-claude-code-vs-cursor-april-2026/
October 16, 2026 — Update (1 page)
Scribbl added as entry 4; Sources section added
- Was: Compared Granola, Fireflies and Otter
- Now: Added Scribbl as entry 4: bot-free capture on Google Meet (Chrome extension) and on Zoom and Teams (Mac/Windows desktop app); free plan 10 meetings a month with no length limit; Pro $13 per user per month billed annually; added a Sources section
- Source: https://scribbl.co/pricing
- Page: /answers/granola-vs-fireflies-vs-otter-april-2026/
October 16, 2026 — Update (1 page)
Design.com AI Image Generator added as entry 6
- Was: Leaderboard covered Midjourney, FLUX, Ideogram, Google Nano Banana/Imagen and GPT Image 2.5 only
- Now: Added entry 6, Design.com AI Image Generator (DesignCrowd Pty Ltd): web-only template-editor generator, no API, no disclosed model or per-image price, account needed to download; one public-demo check on October 16, 2026, not ranked in the head-to-head table
- Source: https://www.design.com/ai-image-generator
- Page: /answers/best-ai-image-generators-june-2026-midjourney-flux-imagen/
October 16, 2026 — Correction (1 page)
Corrected /posts/kimi-k3-in-c-278-trillion-params-8gb-ram-review/
- Was: The whole engine is 176 KB of C across seven source files
- Now: The whole engine compiles to a 176 KB binary (the README gives 176 KB as the binary size; its own C-file counts disagree)
- Why it happened: claim audit
- Source: https://github.com/FareedKhan-dev/kimi-k3-in-c
- Page: /posts/kimi-k3-in-c-278-trillion-params-8gb-ram-review/
October 16, 2026 — Correction (1 page)
Corrected /posts/voicestudio-review-local-elevenlabs-alternative/
- Was: Prebuilt packages: an Apple Silicon DMG, a Windows x64 MSI (with a current-user build), and a Linux AppImage
- Now: Apple Silicon and Intel DMGs, a Windows x64 .exe, and a Linux AppImage or .deb (Electron builds since v0.5.4)
- Why it happened: claim audit
- Source: https://github.com/debpalash/VoiceStudio/releases/tag/v0.5.6
- Page: /posts/voicestudio-review-local-elevenlabs-alternative/
October 16, 2026 — Update (1 page)
Updated /posts/ifixai-review-open-source-ai-agent-auditing/
- Was: 8 open pull requests
- Now: 10 open pull requests (October 16, 2026)
- Why it happened: claim audit
- Source: https://github.com/ifixai-ai/iFixAi
- Page: /posts/ifixai-review-open-source-ai-agent-auditing/
October 16, 2026 — Update (1 page)
Updated /answers/newest-gpt-vs-claude-models-price-and-performance-2026/
- Was: Output speed (Artificial Analysis): Opus 5.5 ~95, GPT-6 Astra ~44, Sonnet 5.5 ~129, GPT-6.1 Sol ~48, Haiku 5.5 ~237, GPT-6 Luna ~116 tok/s
- Now: Opus 5.5 ~96, GPT-6 Astra ~54, Sonnet 5.5 ~135, GPT-6.1 Sol ~54, Haiku 5.5 ~238, GPT-6 Luna ~120 tok/s (read October 16, 2026)
- Why it happened: claim audit
- Source: https://artificialanalysis.ai/leaderboards/models
- Page: /answers/newest-gpt-vs-claude-models-price-and-performance-2026/
October 16, 2026 — Correction (1 page)
Corrected /answers/best-claude-code-alternatives-2026/
- Was: Gemini CLI is free with a personal Google account at 60 requests per minute and 1,000 requests per day
- Now: Gemini CLI stopped serving free personal accounts on June 18, 2026 (moved to Antigravity CLI); it needs a paid Gemini API key or a Code Assist Standard/Enterprise licence
- Why it happened: claim audit
- Source: https://developers.googleblog.com/an-important-update-transitioning-gemini-cli-to-antigravity-cli/
- Page: /answers/best-claude-code-alternatives-2026/
October 15, 2026 — Update (1 page)
Updated /posts/kimi-k3-in-c-278-trillion-params-8gb-ram-review/
- Was: No macOS/Apple Silicon or ARM build; base model with no chat template; latest v1.0.0
- Now: v1.1.0 (October 2, 2026): macOS/arm64 and Windows build and pass every gate with NEON kernels; --chat uses the official K3 XTML format
- Source: https://github.com/FareedKhan-dev/kimi-k3-in-c/releases/tag/v1.1.0
- Page: /posts/kimi-k3-in-c-278-trillion-params-8gb-ram-review/
October 15, 2026 — Update (1 page)
Updated /posts/unity-mcp-ai-game-development-bridge/
- Was: Latest release v10.2.0 (September 1, 2026)
- Now: Latest release v10.3.0 (October 4, 2026)
- Source: https://github.com/CoplayDev/unity-mcp/releases/tag/v10.3.0
- Page: /posts/unity-mcp-ai-game-development-bridge/
October 15, 2026 — Update (1 page)
Updated /posts/agent-reach-ai-agent-internet-cli-review/
- Was: Install with pip install agent-reach; agent-reach install --env=auto installs dependencies
- Now: Install from the GitHub archive (pipx install https://github.com/Panniantong/agent-reach/archive/main.zip) — the PyPI package of the same name is a different project; agent-reach install is check-only unless --system is passed
- Source: https://raw.githubusercontent.com/Panniantong/agent-reach/main/docs/install.md
- Page: /posts/agent-reach-ai-agent-internet-cli-review/
October 15, 2026 — Update (1 page)
Updated /posts/caveman-claude-code-skill-token-savings-review/
- Was: Latest v2.5.0; license MIT with BSL-1.1 for Engine-linked directories; install via install.sh from main
- Now: Latest v4.0.1 (October 13, 2026); whole repository Apache-2.0 from v3.0.0; recommended install npx @caveman-ai/cli@latest
- Source: https://github.com/JuliusBrussee/caveman/blob/main/LICENSING.md
- Page: /posts/caveman-claude-code-skill-token-savings-review/
October 15, 2026 — Update (1 page)
Updated /posts/voicestudio-review-local-elevenlabs-alternative/
- Was: Latest release v0.5.1 (August 28, 2026); Tauri v2 desktop app; 16 TTS and 11 ASR engines
- Now: Latest release v0.5.6 (September 23, 2026); Electron desktop app (v0.5.3 was the final Tauri release); feature catalog lists 17 speech-generation engines and 10 transcription options
- Source: https://github.com/debpalash/VoiceStudio/releases/tag/v0.5.6
- Page: /posts/voicestudio-review-local-elevenlabs-alternative/
October 15, 2026 — Correction (1 page)
Corrected /answers/low-latency-llm-inference-providers-2026/
- Was: Artificial Analysis measures GPT-6 Luna at about 135 output tokens per second
- Now: Artificial Analysis measures GPT-6 Luna at about 116 output tokens per second
- Why it happened: claim audit
- Source: https://artificialanalysis.ai/leaderboards/models
- Page: /answers/low-latency-llm-inference-providers-2026/
October 15, 2026 — Correction (1 page)
Corrected /answers/newest-gpt-vs-claude-models-price-and-performance-2026/
- Was: Output speed: GPT-6 Astra ~60 tok/s, Claude Sonnet 5.5 ~136 tok/s, GPT-6.1 Sol ~59 tok/s, GPT-6 Luna ~135 tok/s
- Now: Output speed: GPT-6 Astra ~44 tok/s, Claude Sonnet 5.5 ~129 tok/s, GPT-6.1 Sol ~48 tok/s, GPT-6 Luna ~116 tok/s (Artificial Analysis, max effort, read October 15, 2026)
- Why it happened: claim audit
- Source: https://artificialanalysis.ai/leaderboards/models
- Page: /answers/newest-gpt-vs-claude-models-price-and-performance-2026/
October 14, 2026 — Correction (1 page)
Corrected /answers/best-llm-inference-providers-2026-ranked/
- Was: Fireworks Series D July 16, 2026
- Now: Fireworks Series D July 15, 2026
- Why it happened: claim audit
- Source: https://fireworks.ai/blog/series-d-announcement
- Page: /answers/best-llm-inference-providers-2026-ranked/
October 14, 2026 — Correction (1 page)
Corrected /answers/what-is-jetbrains-mellum2-1-open-coding-model-2026/
- Was: Mellum2.1 not yet in Ollama or LM Studio; GGUF builds 'coming soon'
- Now: JetBrains publishes GGUF builds of Mellum2.1 (BF16, Q8_0, Q6_K, Q4_K_M 8.1 GB, MXFP4 7.0 GB) for llama.cpp, Ollama and LM Studio
- Why it happened: claim audit
- Source: https://huggingface.co/JetBrains/Mellum2.1-12B-A2.5B-Thinking-GGUF
- Page: /answers/what-is-jetbrains-mellum2-1-open-coding-model-2026/
October 13, 2026 — Update (1 page)
Updated /answers/best-ai-agent-control-planes-may-2026/
- Was: Agent 365 priced via M365 E7 $99/user/mo; Bedrock Managed Agents limited preview; LangGraph Platform
- Now: Agent 365 $15/user/mo standalone or in E7 $99; Bedrock Managed Agents in broader preview (Sep 2026) with no extra charge; LangGraph Platform now LangSmith Deployment (Plus $39/seat); ServiceNow AI Control Tower added
- Source: https://go.microsoft.com/fwlink/?linkid=2359230
- Page: /answers/best-ai-agent-control-planes-may-2026/
October 13, 2026 — Update (1 page)
Updated /answers/best-llm-inference-providers-2026-ranked/
- Was: Baseten $13B valuation; Cerebras powers OpenAI's Ultrafast mode
- Now: Baseten reported in talks at ~$26B after OpenAI partnership (not closed); OpenAI Ultrafast now a paid GPT-6 Astra tier at 6x, hardware partner unconfirmed
- Source: https://openai.com/index/previewing-ultrafast/
- Page: /answers/best-llm-inference-providers-2026-ranked/
October 13, 2026 — Update (1 page)
Updated /answers/v0-vs-lovable-vs-bolt-new-april-2026/
- Was: v0 Pro $20/mo (500 msgs); Bolt Pro $20/mo; Bolt free 1M daily tokens
- Now: v0 $20 Premium closed to new users, Plus $30/user, Business $100/user, token-billed v0 models; Bolt Pro $25/mo, Teams $30/member; Bolt free 300K/day, 1M/month; Lovable Pro $25 for 100 credits/month
- Source: https://v0.app/docs/pricing
- Page: /answers/v0-vs-lovable-vs-bolt-new-april-2026/
October 13, 2026 — Update (1 page)
Updated /answers/is-anthropic-profitable-first-profit-q2-2026-explained/
- Was: OpenAI's S-1 is reported as imminent; Sonnet 5 intro $2/$10 through August 31, 2026
- Now: Altman delayed OpenAI's IPO past 2026; Sonnet 5 $2/$10 is permanent (Sep 1 rise cancelled); Opus 5.5 is $4/$20; Anthropic public S-1 not yet on EDGAR, investor day Oct 14 reported
- Source: https://platform.claude.com/docs/en/about-claude/pricing
- Page: /answers/is-anthropic-profitable-first-profit-q2-2026-explained/
October 13, 2026 — Update (1 page)
Updated /answers/ios-27-liquid-glass-vs-ios-26-accessibility-changes-june-2026/
- Was: General availability expected mid-September 2026
- Now: iOS 27 released September 14, 2026; iOS 27.0.1 on September 28
- Source: https://www.macworld.com/article/2986799/ios-27-features-compatiblity-siri-ai-updates.html
- Page: /answers/ios-27-liquid-glass-vs-ios-26-accessibility-changes-june-2026/
October 13, 2026 — Correction (1 page)
Corrected /answers/ios-27-liquid-glass-vs-ios-26-accessibility-changes-june-2026/
- Was: iOS 27 adds a High Contrast Liquid Glass mode and a UIDesignSystem API that lets third-party apps opt out of Liquid Glass
- Now: No such mode or API shipped; iOS 27 adds a Liquid Glass slider (Settings > Appearance), and the UIDesignRequiresCompatibility opt-out is ignored with the iOS 27 SDK
- Why it happened: Unconfirmed WWDC-week reporting presented as fact
- Source: https://www.macrumors.com/how-to/ios-27-tone-down-liquid-glass-transparency/
- Page: /answers/ios-27-liquid-glass-vs-ios-26-accessibility-changes-june-2026/
October 13, 2026 — Correction (1 page)
Corrected /answers/ios-27-liquid-glass-vs-ios-26-accessibility-changes-june-2026/
- Was: iOS 27 drops iPhone 11, 11 Pro and 11 Pro Max; minimum supported iPhone is iPhone 12
- Now: iOS 27 supports iPhone 11 and iPhone SE (2nd gen) and later; no iPhone was dropped
- Why it happened: Written from WWDC-week reporting before Apple's compatibility list was checked
- Source: https://support.apple.com/guide/iphone/iphone-models-compatible-with-ios-27-iphe3fa5df43/ios
- Page: /answers/ios-27-liquid-glass-vs-ios-26-accessibility-changes-june-2026/
October 12, 2026 — Correction (1 page)
Corrected /answers/best-pdf-parsing-api-for-rag-pipelines-2026/
- Was: LlamaParse outputs Markdown and per-page JSON on all tiers, including Fast
- Now: Markdown and per-page JSON on the Cost-effective tier and up; the Fast tier outputs spatial text only, no Markdown
- Why it happened: claim audit
- Source: https://developers.llamaindex.ai/llamaparse/general/pricing/
- Page: /answers/best-pdf-parsing-api-for-rag-pipelines-2026/
October 12, 2026 — Correction (1 page)
Corrected /answers/alternatives-to-dosu-ai-memory-knowledge-systems-2026/
- Was: Dosu auto-labels issues and pull requests as a hosted feature (comparison table: Issue triage / labelling: Yes)
- Now: Auto-labeling is no longer a hosted Dosu feature; Dosu open-sourced it as auto-label (and better-stale-bot), which run in your own GitHub Actions
- Why it happened: claim audit
- Source: https://dosu.dev/pricing
- Page: /answers/alternatives-to-dosu-ai-memory-knowledge-systems-2026/
October 11, 2026 — Correction (1 page)
Corrected /answers/best-arm-laptops-for-ai-llm-inference-2026/
- Was: Surface Laptop Ultra from $2,599.99
- Now: Surface Laptop Ultra from $2,599 (Microsoft's stated MSRP)
- Why it happened: claim audit
- Source: https://blogs.windows.com/devices/2026/10/07/pre-order-our-most-powerful-surface-devices-ever/
- Page: /answers/best-arm-laptops-for-ai-llm-inference-2026/
October 10, 2026 — Correction (1 page)
Corrected /answers/best-ocr-document-extraction-apis-2026-ranked/
- Was: Google Document AI prebuilt processors (invoice, receipt, expense, W-2, bank statement, ID) cost $10 per 1,000 pages
- Now: Google's Layout Parser is $10 per 1,000 pages; its prebuilt invoice, expense and ID parsers are $0.10 per document of up to 10 pages (W-2 $0.30, bank statement $0.75 per document)
- Why it happened: claim audit
- Source: https://cloud.google.com/document-ai/pricing
- Page: /answers/best-ocr-document-extraction-apis-2026-ranked/
October 9, 2026 — Update (1 page)
Updated /answers/nemotron-3-nano-omni-vs-llama-5-vs-qwen-3-6-multimodal-april-2026/
- Was: Qwen 3.7 is rumored
- Now: Qwen 3.8 shipped in August 2026 (27B Apache-2.0 on Aug 5; 2.4T Qwen3.8-Max on Aug 8)
- Source: https://huggingface.co/Qwen/Qwen3.8-27B
- Page: /answers/nemotron-3-nano-omni-vs-llama-5-vs-qwen-3-6-multimodal-april-2026/
October 9, 2026 — Correction (1 page)
Corrected /answers/nemotron-3-nano-omni-vs-llama-5-vs-qwen-3-6-multimodal-april-2026/
- Was: Qwen 3.6 ships 7B, 72B and 235B MoE variants; multimodality needs separate Qwen-VL models
- Now: Qwen 3.6 open weights are 35B-A3B and 27B (Apache-2.0), both natively image+text
- Source: https://huggingface.co/Qwen/Qwen3.6-35B-A3B
- Page: /answers/nemotron-3-nano-omni-vs-llama-5-vs-qwen-3-6-multimodal-april-2026/
October 9, 2026 — Correction (1 page)
Corrected /answers/nemotron-3-nano-omni-vs-llama-5-vs-qwen-3-6-multimodal-april-2026/
- Was: Nemotron 3 Super and Ultra are coming in the first half of 2026
- Now: Nemotron 3 Super (120B-A12B) was released March 2026, before publication; Ultra (550B-A55B) followed June 4, 2026
- Source: https://research.nvidia.com/labs/nemotron/Nemotron-3-Ultra/
- Page: /answers/nemotron-3-nano-omni-vs-llama-5-vs-qwen-3-6-multimodal-april-2026/
October 9, 2026 — Correction (1 page)
Corrected /answers/nemotron-3-nano-omni-vs-llama-5-vs-qwen-3-6-multimodal-april-2026/
- Was: Nemotron 3 Nano Omni has ~8-12B active parameters
- Now: Nemotron 3 Nano Omni has ~3B active parameters (30B total), 256K context
- Source: https://huggingface.co/nvidia/Nemotron-3-Nano-Omni-30B-A3B-Reasoning-BF16
- Page: /answers/nemotron-3-nano-omni-vs-llama-5-vs-qwen-3-6-multimodal-april-2026/
October 9, 2026 — Update (1 page)
Updated /answers/best-ocr-document-extraction-apis-2026-ranked/
- Was: Mistral OCR 4 (mistral-ocr-4-0) is the current model; OCR 4.1 in public preview
- Now: OCR 4.1 GA since August 31, 2026; OCR 4.0 retired September 30, 2026; same price ($4 per 1,000 pages, $2 batch)
- Source: https://docs.mistral.ai/resources/changelogs
- Page: /answers/best-ocr-document-extraction-apis-2026-ranked/
October 9, 2026 — Update (1 page)
Updated /answers/ios-27-siri-extensions-hidden-claude-chatgpt-gemini-june-2026/
- Was: iOS 27 public release expected September 2026; Claude/Gemini Siri Extensions may ship in 27.0 or 27.1/27.2
- Now: iOS 27 shipped September 14, 2026 with Siri AI (beta, English, not initially in the EU) but without Claude or Gemini extensions; ChatGPT remains the only third-party model in Siri
- Source: https://www.apple.com/newsroom/2026/09/siri-ai-a-profoundly-more-capable-and-personal-assistant-is-here/
- Page: /answers/ios-27-siri-extensions-hidden-claude-chatgpt-gemini-june-2026/
October 9, 2026 — Correction (1 page)
Corrected /answers/mastra-vs-langchain-vs-vercel-ai-sdk-typescript-agent-frameworks-2026/
- Was: LangChain has no native guardrails (external only)
- Now: LangChain.js 1.0+ implements guardrails (PII, prompt-injection, output validation) via agent middleware
- Source: https://docs.langchain.com/oss/javascript/langchain/guardrails
- Page: /answers/mastra-vs-langchain-vs-vercel-ai-sdk-typescript-agent-frameworks-2026/
October 9, 2026 — Correction (1 page)
Corrected /answers/mastra-vs-langchain-vs-vercel-ai-sdk-typescript-agent-frameworks-2026/
- Was: Mastra launched January 2026
- Now: Mastra was first released in October 2024; version 1.0 shipped January 20, 2026
- Source: https://mastra.ai/blog/changelog-2026-01-20
- Page: /answers/mastra-vs-langchain-vs-vercel-ai-sdk-typescript-agent-frameworks-2026/
October 9, 2026 — Correction (1 page)
Corrected /answers/mastra-vs-langchain-vs-vercel-ai-sdk-typescript-agent-frameworks-2026/
- Was: Mastra and the Vercel AI SDK are MIT-licensed
- Now: Mastra and the Vercel AI SDK are Apache-2.0 (Mastra's ee/ directories separately licensed); LangChain is MIT
- Source: https://github.com/mastra-ai/mastra/blob/main/LICENSE.md
- Page: /answers/mastra-vs-langchain-vs-vercel-ai-sdk-typescript-agent-frameworks-2026/
October 9, 2026 — Correction (1 page)
Corrected /answers/best-ai-image-generators-june-2026-midjourney-flux-imagen/
- Was: Stable Diffusion 4 (SD4) recommended as the best self-hosted option
- Now: Stability AI has not released an SD4; its newest Stable Diffusion image model is SD 3.5. Self-hosted pick is open-weight FLUX.2 [dev]/[klein] or SD 3.5
- Source: https://stability.ai/news-updates
- Page: /answers/best-ai-image-generators-june-2026-midjourney-flux-imagen/
October 9, 2026 — Update (1 page)
Updated /answers/best-ai-image-generators-june-2026-midjourney-flux-imagen/
- Was: Ideogram v3; plans $7 Basic / $16 Plus / $48 Pro per month
- Now: Ideogram 4.5 is the newest model; Basic no longer sold; Plus $20/month ($15 yearly), Pro $60/month ($42 yearly), Team $30 per user
- Source: https://docs.ideogram.ai/plans-and-billing/available-plans
- Page: /answers/best-ai-image-generators-june-2026-midjourney-flux-imagen/
October 9, 2026 — Update (1 page)
Updated /answers/best-ai-image-generators-june-2026-midjourney-flux-imagen/
- Was: Imagen 4 is Google's image pick at $0.04/image
- Now: Google's current image models are Nano Banana 2.1 ($0.0336 per 1K image) and Nano Banana Pro ($0.134 per 1K image); Imagen 4 price last confirmed June 7, 2026
- Source: https://the-decoder.com/googles-new-image-model-nano-banana-2-1-generates-better-images-for-less-money/
- Page: /answers/best-ai-image-generators-june-2026-midjourney-flux-imagen/
October 9, 2026 — Update (1 page)
Updated /answers/best-ai-image-generators-june-2026-midjourney-flux-imagen/
- Was: GPT Image 2 at ~$0.07 standard / $0.13 HD per image
- Now: GPT Image 2.5 (Flare and Sunburst) is current; token-based $8/M image input, $30/M image output, $5/M text input — about $0.053 per 1024x1024 image at high quality
- Source: https://developers.openai.com/api/docs/pricing
- Page: /answers/best-ai-image-generators-june-2026-midjourney-flux-imagen/
October 9, 2026 — Update (1 page)
Updated /answers/best-ai-image-generators-june-2026-midjourney-flux-imagen/
- Was: Flux 1.1 Pro at $0.04/image is Black Forest Labs' top API model
- Now: FLUX 3 Image (launched October 1, 2026) costs $0.048 per 1K image; FLUX.2 [pro] from $0.03; FLUX1.1 [pro] is previous generation at $0.04
- Source: https://docs.bfl.ml/quick_start/pricing
- Page: /answers/best-ai-image-generators-june-2026-midjourney-flux-imagen/
October 9, 2026 — Update (1 page)
Updated /answers/best-ai-image-generators-june-2026-midjourney-flux-imagen/
- Was: Midjourney v7 (v7.2 March 2026) is the current Midjourney model
- Now: Midjourney V8.2 (released July 24, 2026) is the current model; plans unchanged at $10/$30/$60/$120 a month, still no API
- Source: https://updates.midjourney.com/version-8-2/
- Page: /answers/best-ai-image-generators-june-2026-midjourney-flux-imagen/
October 9, 2026 — Update (1 page)
Updated /answers/best-inference-api-vendors-autoscaling-fast-cold-start-2026/
- Was: Dedicated RunPod H100 SXM pod $3.49/h vs $4.79 serverless, a 37% premium; dedicated wins above ~75% utilisation
- Now: Dedicated RunPod H100 SXM pod $3.99/h vs $4.79 serverless, a 20% premium; dedicated wins above ~80% utilisation
- Why it happened: claim audit: RunPod raised list prices after publication
- Source: https://www.runpod.io/pricing
- Page: /answers/best-inference-api-vendors-autoscaling-fast-cold-start-2026/
October 9, 2026 — Update (1 page)
Updated /answers/how-to-self-host-an-llm-for-production-2026-guide/
- Was: RunPod on-demand: L4 $0.49, H100 SXM $3.49, RTX Pro 6000 $2.09, H200 $4.59, B200 $6.79 per hour; one H100 about $2,550 a month
- Now: RunPod on-demand (Secure Cloud): L4 $0.59, H100 SXM $3.99, RTX Pro 6000 $2.49, H200 $5.29, B200 $7.99 per hour; one H100 about $2,900 a month
- Why it happened: claim audit: RunPod raised list prices after publication (Wayback 2026-10-08 showed the old rates)
- Source: https://www.runpod.io/pricing
- Page: /answers/how-to-self-host-an-llm-for-production-2026-guide/
October 8, 2026 — Update (1 page)
Updated /posts/polsia-1m-arr-30-days-zero-employees/
- Was: Live feed: 19,296 active of 318,085 companies; $10.2M ARR ($7.0M recurring); Trustpilot 1.8/5 (79 reviews)
- Now: Live feed Sep 15, 2026: 29,132 active of 484,592; $18.7M ARR ($11.0M recurring); Trustpilot 3.3/5 (347 reviews) on Oct 2, 2026
- Source: https://preuve.ai/blog/polsia-review
- Page: /posts/polsia-1m-arr-30-days-zero-employees/
October 8, 2026 — Update (1 page)
Updated /posts/polsia-1m-arr-30-days-zero-employees/
- Was: Polsia takes a 20% revenue share on user businesses
- Now: Polsia charges a 3% platform fee on customer payments with a 31-day settlement hold, plus a 20% fee on ad spend (Terms of Sep 17, 2026)
- Source: https://polsia.com/terms
- Page: /posts/polsia-1m-arr-30-days-zero-employees/
October 8, 2026 — Update (1 page)
Updated /posts/browser-use-ai-agent-browser-automation/
- Was: CLI used open/state/click commands; cloud profile sync via browser-use.com/profile.sh
- Now: CLI is backed by Browser Harness and takes Python (new_tab, page_info); profile.sh returns 404, profile sync follows the Browser Harness guide
- Source: https://docs.browser-use.com/open-source/browser-use-cli
- Page: /posts/browser-use-ai-agent-browser-automation/
October 8, 2026 — Correction (1 page)
Corrected /posts/browser-use-ai-agent-browser-automation/
- Was: Model examples imported ChatOpenAI/ChatAnthropic/ChatOllama from LangChain packages; Chrome profile via BrowserConfig(chrome_instance_path=...)
- Now: Model classes are imported from browser_use directly; reuse a Chrome profile with Browser.from_system_chrome()
- Source: https://github.com/browser-use/browser-use
- Page: /posts/browser-use-ai-agent-browser-automation/
October 8, 2026 — Update (1 page)
Updated /posts/browser-use-ai-agent-browser-automation/
- Was: ChatBrowserUse priced at $0.20 input / $0.02 cached / $2.00 output per 1M tokens; new accounts get $10 free credits
- Now: BU2 (bu-2-0) priced at $0.60 input / $0.06 cached / $3.50 output per 1M tokens; signup credit is one-time and varies; cloud browsers $0.02 per browser-hour
- Source: https://docs.browser-use.com/open-source/supported-models
- Page: /posts/browser-use-ai-agent-browser-automation/
October 8, 2026 — Correction (1 page)
Corrected /posts/awesome-llm-apps-deep-dive/
- Was: Getting-started example ran the AI Deep Research Agent with 'python main.py'
- Now: The agent is a Streamlit app started with 'streamlit run deep_research_openai.py'
- Source: https://github.com/Shubhamsaboo/awesome-llm-apps/tree/main/advanced_ai_agents/single_agent_apps/ai_deep_research_agent
- Page: /posts/awesome-llm-apps-deep-dive/
October 8, 2026 — Update (1 page)
Updated /posts/ollama-mlx-apple-silicon-review/
- Was: Latest Ollama release v0.32.1; MLX stable since v0.30
- Now: Latest release v0.40.1 (Oct 7, 2026); MLX runs supported models by default on Apple Silicon since v0.40.0 (Sep 25, 2026)
- Source: https://github.com/ollama/ollama/releases/tag/v0.40.0
- Page: /posts/ollama-mlx-apple-silicon-review/
October 8, 2026 — Correction (1 page)
Corrected /answers/best-enterprise-ai-ocr-document-processing-vendors-2026/
- Was: Azure's price list starts at $12,960 a year for 2 million Read pages
- Now: Azure's price list starts at $12,960 a year for up to 24 million Read pages, about 2 million a month
- Why it happened: claim audit
- Source: https://azure.microsoft.com/en-us/pricing/details/document-intelligence/
- Page: /answers/best-enterprise-ai-ocr-document-processing-vendors-2026/
October 8, 2026 — Correction (1 page)
Corrected /posts/kev-review-open-source-jev-decision-models-qwen/
- Was: Jared Palmer, Vercel's VP of AI
- Now: Jared Palmer, VP of Engineering at Cognition (the company behind Devin)
- Why it happened: claim audit
- Source: https://jaredpalmer.com/blog/introducing-kev
- Page: /posts/kev-review-open-source-jev-decision-models-qwen/
October 7, 2026 — Update (1 page)
Updated /reference/ai-model-pricing/
- Was: DeepSeek V4 Flash: $0.22 input / $0.66 output per 1M tokens (off-peak)
- Now: DeepSeek V4 Flash is retired; the deepseek-v4-flash name is served by DeepSeek-V4.1-Flash and billed $0.15 input / $0.60 output per 1M tokens (off-peak)
- Why it happened: claim audit
- Source: https://api-docs.deepseek.com/quick_start/pricing
- Page: /reference/ai-model-pricing/
October 7, 2026 — Correction (1 page)
Corrected /answers/what-hardware-do-i-need-to-run-llms-locally-2026/
- Was: 64 GB Mac mini M5 Pro ($2,699)
- Now: 64 GB Mac mini M5 Pro ($3,199 with 1 TB on apple.com)
- Why it happened: claim audit
- Source: https://www.apple.com/shop/buy-mac/mac-mini/m5-pro-chip-18-core-cpu-20-core-gpu-64gb-memory-1tb-storage
- Page: /answers/what-hardware-do-i-need-to-run-llms-locally-2026/
October 7, 2026 — Correction (1 page)
Corrected /answers/microsoft-agent-365-vs-google-adk-vs-aws-bedrock-agentcore-may-2026/
- Was: AWS made AgentCore Harness generally available (July 2026)
- Now: AWS made AgentCore Harness generally available (June 2026)
- Why it happened: claim audit
- Source: https://docs.aws.amazon.com/bedrock-agentcore/latest/devguide/release-notes.html
- Page: /answers/microsoft-agent-365-vs-google-adk-vs-aws-bedrock-agentcore-may-2026/
October 7, 2026 — Correction (1 page)
Corrected /answers/baseten-vs-together-ai-vs-fireworks-ai-inference-platforms-june-2026/
- Was: Baseten global clusters: 87 (as of June 2026)
- Now: Baseten global clusters: 90+ across 20+ clouds (September 2026)
- Why it happened: claim audit
- Source: https://www.baseten.co/blog/baseten-openai-partnership/
- Page: /answers/baseten-vs-together-ai-vs-fireworks-ai-inference-platforms-june-2026/
October 7, 2026 — Correction (1 page)
Corrected /answers/baseten-vs-together-ai-vs-fireworks-ai-inference-platforms-june-2026/
- Was: Fireworks AI raised a $1.505B Series D on July 16, 2026
- Now: Fireworks AI raised a $1.505B Series D on July 15, 2026
- Why it happened: claim audit
- Source: https://fireworks.ai/blog/series-d-announcement
- Page: /answers/baseten-vs-together-ai-vs-fireworks-ai-inference-platforms-june-2026/
October 7, 2026 — Correction (1 page)
Corrected /answers/how-to-link-claude-code-to-unity-mcp-2026/
- Was: CoplayDev's MCP for Unity exposes 48 tool entrypoints
- Now: CoplayDev's MCP for Unity exposes 50 tool entrypoints
- Why it happened: claim audit
- Source: https://github.com/CoplayDev/unity-mcp
- Page: /answers/how-to-link-claude-code-to-unity-mcp-2026/
October 6, 2026 — Correction (1 page)
Corrected /answers/microsoft-agent-365-vs-google-adk-vs-aws-bedrock-agentcore-may-2026/
- Was: Google ADK launched in 2026; Gemini 4 available as an ADK model
- Now: ADK was first released in April 2025 (now v2.11.0); Gemini 4 Argon was announced Sep 30, 2026 and is not generally available
- Source: https://github.com/google/adk-python
- Page: /answers/microsoft-agent-365-vs-google-adk-vs-aws-bedrock-agentcore-may-2026/
October 6, 2026 — Correction (1 page)
Corrected /answers/microsoft-agent-365-vs-google-adk-vs-aws-bedrock-agentcore-may-2026/
- Was: E7 Frontier Suite priced at $XX/seat; Agent 365 price not stated
- Now: Agent 365 $15/user/month standalone; Microsoft 365 E7 $99/user/month (both GA May 1, 2026)
- Source: https://blogs.microsoft.com/blog/2026/03/09/introducing-the-first-frontier-suite-built-on-intelligence-trust/
- Page: /answers/microsoft-agent-365-vs-google-adk-vs-aws-bedrock-agentcore-may-2026/
October 6, 2026 — Update (1 page)
Updated /answers/baseten-vs-together-ai-vs-fireworks-ai-inference-platforms-june-2026/
- Was: Per-token price ranges for Llama 4 70B / Qwen 3.5 / DeepSeek V3.5 (~$0.27-$1.00 in)
- Now: October 2026 list prices for Kimi K3, Qwen3.8 Max, GLM-5.3, DeepSeek V4.1 Flash, MiniMax M3 on Together and Fireworks
- Source: https://www.together.ai/pricing
- Page: /answers/baseten-vs-together-ai-vs-fireworks-ai-inference-platforms-june-2026/
October 6, 2026 — Update (1 page)
Updated /answers/baseten-vs-together-ai-vs-fireworks-ai-inference-platforms-june-2026/
- Was: Together AI: prior raises ($2B+ valuation tier); Fireworks AI: prior raises ($500M-$1B+ tier)
- Now: Together AI: $800M Series C at $8.3B (July 1, 2026); Fireworks AI: $1.505B Series D at $17.5B (July 16, 2026), $1B+ ARR
- Source: https://fireworks.ai/blog/series-d-announcement
- Page: /answers/baseten-vs-together-ai-vs-fireworks-ai-inference-platforms-june-2026/
October 6, 2026 — Update (1 page)
Updated /answers/ai-coding-agent-market-share-cursor-claude-code-copilot-july-2026/
- Was: Cursor valuation $29.3B (Nov 2025)
- Now: SpaceX completed a $60B all-stock acquisition of Cursor's parent Anysphere on August 14, 2026
- Source: https://seekingalpha.com/news/4633335-spacex-completes-60b-acquisition-of-cursor-as-musk-led-firm-tries-to-gain-edge-in-ai-coding
- Page: /answers/ai-coding-agent-market-share-cursor-claude-code-copilot-july-2026/
October 6, 2026 — Correction (1 page)
Corrected /answers/ai-coding-agent-market-share-cursor-claude-code-copilot-july-2026/
- Was: Cursor ARR $2B+ (Feb 2026)
- Now: Cursor reached a $3B revenue run rate in May 2026 (Bloomberg)
- Source: https://en.wikipedia.org/wiki/Cursor_(company)
- Page: /answers/ai-coding-agent-market-share-cursor-claude-code-copilot-july-2026/
October 6, 2026 — Update (1 page)
Updated /answers/ai-coding-agent-market-share-cursor-claude-code-copilot-july-2026/
- Was: GitHub Copilot has ~20M total users
- Now: GitHub Copilot has 50 million users (Nadella, July 29, 2026 earnings call)
- Source: https://www.cnbc.com/2026/07/29/microsoft-msft-q4-earnings-report-2026.html
- Page: /answers/ai-coding-agent-market-share-cursor-claude-code-copilot-july-2026/
October 6, 2026 — Correction (1 page)
Corrected /answers/kilo-code-vs-cline-vs-continue-open-source-coding-agents-july-2026/
- Was: Recommended Continue for team and JetBrains use
- Now: Continue made its final 2.0.0 release in June 2026 and is no longer maintained; not recommended for new setups
- Source: https://github.com/continuedev/continue
- Page: /answers/kilo-code-vs-cline-vs-continue-open-source-coding-agents-july-2026/
October 6, 2026 — Correction (1 page)
Corrected /answers/kilo-code-vs-cline-vs-continue-open-source-coding-agents-july-2026/
- Was: Kilo Code is Apache 2.0 licensed and was forked from Roo Code in May 2026
- Now: Kilo Code is MIT-licensed and its repository dates from March 2025
- Source: https://github.com/Kilo-Org/kilocode
- Page: /answers/kilo-code-vs-cline-vs-continue-open-source-coding-agents-july-2026/
October 6, 2026 — Correction (1 page)
Corrected /answers/what-is-gemini-4-google-io-may-2026/
- Was: Gemini 4 was unveiled at the Google I/O keynote on May 19, 2026 and went live in the Gemini app, Search AI Mode and Vertex AI that day, with a 10M-token context window
- Now: Google I/O 2026 launched the Gemini 3.5 series, not Gemini 4. Gemini 4 Argon was announced on September 30, 2026, in limited release (Fairwind Program); the 10M context claim was never confirmed
- Source: https://blog.google/innovation-and-ai/models-and-research/gemini-models/gemini-4-argon/
- Page: /answers/what-is-gemini-4-google-io-may-2026/
October 6, 2026 — Correction (1 page)
Corrected /answers/do-claude-console-api-credits-transfer-to-claude-app-2026/
- Was: Contact Anthropic support through the help centre about refunds rather than expecting a transfer.
- Now: Anthropic's help centre states that all credit purchases are non-refundable, so expect neither a refund nor a transfer.
- Why it happened: claim audit
- Source: https://support.claude.com/en/articles/8977456-how-do-i-pay-for-my-claude-api-usage
- Page: /answers/do-claude-console-api-credits-transfer-to-claude-app-2026/
October 6, 2026 — Correction (1 page)
Corrected /answers/best-ocr-pipeline-for-large-scale-data-extraction-2026/
- Was: Datalab reports that Chandra 2 processes about 2 pages per second on one H100 with 96 concurrent requests
- Now: Datalab measured Chandra 2 at 1.44 pages per second on one H100 (vLLM, 96 concurrent sequences) on a hard benchmark set and estimates about 2 pages per second on real-world documents
- Why it happened: claim audit
- Source: https://github.com/datalab-to/chandra
- Page: /answers/best-ocr-pipeline-for-large-scale-data-extraction-2026/
October 5, 2026 — Update (1 page)
Updated /reference/ai-model-pricing/
- Was: Gemini 3.6 Flash $1.50 input / $7.50 output per million tokens
- Now: Gemini 3.6 Flash $0.75 / $3.75 through December 31, 2026 (intro rate), then $1.50 / $7.50 from January 1, 2027
- Why it happened: claim audit
- Source: https://ai.google.dev/gemini-api/docs/pricing
- Page: /reference/ai-model-pricing/
October 5, 2026 — Update (1 page)
Updated /answers/agent-memory-vs-documentation-markdown-workspace-vs-vector-memory-2026/
- Was: Hacker News discussion at 57 points, 42 comments
- Now: 353 points and 237 comments as of October 5, 2026
- Why it happened: claim audit
- Source: https://news.ycombinator.com/item?id=49945933
- Page: /answers/agent-memory-vs-documentation-markdown-workspace-vs-vector-memory-2026/
October 5, 2026 — Correction (1 page)
Corrected /answers/cactus-whistle-vs-whisper-base-vs-moonshine-tiny-on-device-speech-to-text-2026/
- Was: Moonshine Tiny v2 released 2025
- Now: Moonshine Tiny v2 released February 2026
- Why it happened: claim audit
- Source: https://arxiv.org/abs/2602.12241
- Page: /answers/cactus-whistle-vs-whisper-base-vs-moonshine-tiny-on-device-speech-to-text-2026/
October 5, 2026 — Correction (1 page)
Corrected /posts/paperclip-review-open-source-ai-agent-company-orchestration/
- Was: 6.3K open issues
- Now: 2.7K open issues plus 3.6K open pull requests (GitHub's 6.3K figure counts both)
- Why it happened: claim audit
- Source: https://github.com/paperclipai/paperclip
- Page: /posts/paperclip-review-open-source-ai-agent-company-orchestration/
October 5, 2026 — Correction (1 page)
Corrected /answers/meta-muse-spark-solves-open-math-problems-six-papers-explained-october-2026/
- Was: Meta published the six papers on October 3, 2026
- Now: Meta published the six papers on October 2, 2026
- Why it happened: claim audit
- Source: https://research.meta.ai/blog/solving-open-research-problems-together
- Page: /answers/meta-muse-spark-solves-open-math-problems-six-papers-explained-october-2026/
August 6, 2026 — Correction (2 posts)
Removed two duplicate tool reviews (Cognee, published August 2, 2026, and Juggler, published August 6, 2026) that repeated reviews already on the site under near-identical titles.
- Was: Two tools each had two separate review posts.
- Now: One review per tool; the duplicate posts were deleted.
- Why it happened: The topic-selection step had no deterministic duplicate check; a 40-character truncation in the tracker let a repeat through.
- What changed in the process: A mandatory pre-publish dedup script (same GitHub repository = duplicate) now runs before any research starts.
July 22, 2026 — Correction (116 pages)
Corrected Claude API prices across 116 answer pages and posts.
- Was: Claude Opus 4.5–4.8 listed at $15/$75 per million tokens; Claude Fable 5 at $15/$75 on 7 pages; Fast Mode described as a discount on 5 pages.
- Now: Claude Opus 4.5–4.8 at $5/$25 per million tokens (the price since November 2025); Claude Fable 5 at $10/$50; Fast Mode is a 2× premium. Every derived cost-per-task figure on those pages was recomputed and each page re-dated.
- Why it happened: New pages took prices from our own earlier pages instead of the vendor's pricing page, so one stale figure propagated.
- What changed in the process: Prices now come only from a canonical, dated reference checked against vendor pricing pages; our own pages are never accepted as a source for a fact.