AI model API pricing
List prices for 21 current AI models, in US dollars per million tokens, each with the date it was last checked against the vendor's own pricing page. Most recent verification: October 1, 2026. The last column is the cost of one reference task — 30,000 input tokens and 5,000 output tokens — so models can be compared on one number.
| Model | Vendor | Input / MTok | Output / MTok | 30K-in / 5K-out task | Released | Verified |
|---|---|---|---|---|---|---|
| Gemini 4 Argon introductory or promotional rate; not publicly available yet | $2 | $10 | $0.110 | Sep 30, 2026 | 2026-10-01 | |
| Claude Sonnet 5.5 | Anthropic | $2 | $10 | $0.110 | Sep 28, 2026 | 2026-09-29 |
| Claude Opus 5.5 | Anthropic | $4 | $20 | $0.220 | Sep 22, 2026 | 2026-09-23 |
| GPT-6.1 Sol | OpenAI | $2 | $10 | $0.110 | Sep 29, 2026 | 2026-09-30 |
| GPT-6 Sol | OpenAI | $2 | $10 | $0.110 | Sep 22, 2026 | 2026-09-23 |
| GPT-6 Luna | OpenAI | $0.10 | $0.50 | $0.0055 | Sep 22, 2026 | 2026-09-23 |
| MiMo-V2.6-Pro open weights | Xiaomi | $0.435 | $0.87 | $0.017 | Sep 21, 2026 | 2026-09-23 |
| MiMo-V2.6-Flash | Xiaomi | $0.14 | $0.28 | $0.0056 | Sep 21, 2026 | 2026-09-23 |
| GPT-6 Astra | OpenAI | $10 | $50 | $0.550 | Sep 3, 2026 | 2026-09-04 |
| Claude Fable 5.1 | Anthropic | $10 | $50 | $0.550 | Sep 3, 2026 | 2026-09-04 |
| Muse Spark 1.3 | Meta | $1.25 | $4.25 | $0.059 | Sep 2, 2026 | 2026-09-04 |
| Gemini 3.8 Flash introductory or promotional rate | $0.75 | $3.75 | $0.041 | Sep 2, 2026 | 2026-09-04 | |
| Claude Sonnet 5 | Anthropic | $2 | $10 | $0.110 | — | 2026-08-17 |
| GPT-5.6 Sol introductory or promotional rate | OpenAI | $4 | $20 | $0.220 | — | 2026-08-25 |
| Gemini 3.7 Flash introductory or promotional rate | $0.75 | $3.75 | $0.041 | Aug 13, 2026 | 2026-08-16 | |
| Grok 4.7 | xAI | $2 | $6 | $0.090 | Sep 21, 2026 | 2026-09-22 |
| DeepSeek V4.1 Flash off-peak rate; peak hours cost 2×; open weights | DeepSeek | $0.15 | $0.60 | $0.0075 | Sep 10, 2026 | 2026-09-24 |
| DeepSeek V4 Pro off-peak rate; peak hours cost 2× | DeepSeek | $0.66 | $1.98 | $0.030 | Aug 13, 2026 | 2026-08-17 |
| GLM-5.3 | Z.ai | $1.40 | $4.40 | $0.064 | Aug 14, 2026 | 2026-08-15 |
| GLM-5.3 Flash | Z.ai | $0.15 | $0.50 | $0.0070 | Aug 26, 2026 | 2026-09-05 |
| Qwen 3.8 Flash open weights | Alibaba | $0.15 | $0.47 | $0.0069 | Aug 24, 2026 | 2026-09-04 |
How to read this table
- List price, standard tier. Prices are the vendor's published per-token rates for the standard API. Batch, cached-input, fast-mode, long-context and regional rates differ and are not shown; where a row is an introductory, promotional or off-peak rate, it says so.
- Each row has its own verification date. That is the day the price was checked against the vendor's pricing page. Vendors change prices without notice — confirm on the vendor's page (linked in the Vendor column where available) before committing a budget.
- Token price is not task cost. Models differ several-fold in how many tokens they consume to finish the same task, so a cheaper token can produce a more expensive job. The reference-task column removes the input/output mix from the comparison, not the efficiency difference.
- Only individually verified rows are listed. 17 older models whose prices have not been re-checked recently are left out until they are.
- Machine-readable copy: /reference/ai-model-pricing.json.
How this page is maintained
The table is generated from the dated price reference that every andrew.ooo page draws from. That reference is updated whenever our research finds a vendor price change, and each entry is stamped with the date it was checked against the vendor's own pricing page. We never use our own earlier pages as the source for a price — see the editorial policy and the corrections log for why.
Frequently asked
How much do AI model APIs cost per million tokens?
As of October 1, 2026, list prices range from $0.10 to $10 per million input tokens and $0.28 to $50 per million output tokens across the 21 models tracked on this page. Each row shows the date it was checked against the vendor's own pricing page.
Which AI model API is cheapest?
By list price, GPT-6 Luna (OpenAI) is the cheapest model on this page as of October 1, 2026: $0.10 input and $0.50 output per million tokens, about $0.0055 for a 30,000-token-in, 5,000-token-out task. List price is not total cost: models differ several-fold in how many tokens they use to finish the same task.
How current are these prices?
Every row carries its own verification date. The page is rebuilt whenever the underlying reference changes; the most recent verification is October 1, 2026. Vendors change prices without notice, so confirm on the vendor's pricing page before committing a budget.