AI agents · OpenClaw · self-hosting · automation

Quick Answer

ChatGPT Plus vs Pro $100 vs Pro $200 During the Astra Pause

Published:

The short answer

Buy Pro $100 if you need GPT-6 Astra now; stay on Plus if you do not; and get on the waitlist mindset for Pro $200, because OpenAI is not selling it. Since September 10, 2026, OpenAI has paused new sign-ups and upgrades to the $200 “Pro 20x” plan because Astra demand exceeded capacity. Here is how the three consumer paid tiers compare as of September 15, 2026.

PlusPro $100Pro $200
Price$20/month$100/month$200/month
Usage allowanceBaseline5x Plus20x Plus
Flagship modelGPT-5.6 Sol (Astra limited by capacity)GPT-6 Astra + GPT-5.6 familyGPT-6 Astra + GPT-5.6 family
Computer use (Astra)LimitedYesYes
Reasoning effortStandardUp to maxUp to max
New sign-ups (Sep 15, 2026)✅ Open✅ Open❌ Paused since Sep 10
Existing subscribersUnaffectedUnaffectedUnaffected; renewals continue
Promo redemptionsOpenOpenPaused
Best forChat, writing, everyday researchProfessionals running Astra dailyHeavy agentic/coding users (when it reopens)

What happened, in order

  • September 3, 2026: OpenAI releases GPT-6 Astra to a handful of enterprise customers under Trusted Access, then to API and paid plans in stages. Greg Brockman calls it the start of the “AGI era”; its marquee feature is computer use — operating a desktop, filling forms and navigating web pages at “superhuman speed.”
  • September 9: Sottiaux posts that Astra demand is “unprecedented” and warns OpenAI “might have to pause new Pro subscriptions for a bit.”
  • September 10: OpenAI updates its Pro tiers help article: new sign-ups, upgrades and promo redemptions for Pro $200 are paused. Sottiaux: “We wanted to take the smallest step that allows us to continue giving the broadest access possible.” Pro $100, Plus, Go and the API stay open.
  • September 15: still paused. A Polymarket market is now pricing when U.S. sign-ups resume.

The precedent is November 2023, when OpenAI paused ChatGPT Plus sign-ups after its first DevDay. Then, as now, the cause was demand outrunning GPUs — despite OpenAI’s compute rising from 0.2 GW in 2023 to about 1.9 GW in 2025, and new capacity from NVIDIA (10 GW+), AMD (6 GW) and Broadcom (10 GW of custom accelerators) scheduled from the second half of 2026.

Why Astra strains the system more than Sol

Two properties of GPT-6 Astra make a 20x-usage tier expensive to serve:

  1. Computer use is token-hungry. Each step of an agentic desktop task sends screenshots and receives actions; a single “book my travel” task can consume what a hundred chat turns would.
  2. Effort levels. Astra exposes low → medium → high → xhigh → max reasoning effort. Pro users default toward the top, and OpenAI says Astra burns usage allowances faster than GPT-5.6 Sol.

Pricing signals the same thing: on the API Astra is $10/$50 per million tokens, versus $4/$20 for GPT-5.6 Sol after its August 21 cut. A user on a $200 plan running Astra at max effort all day is the most expensive customer OpenAI has.

What to buy, by use case

You want to…ChooseNotes
Chat, draft, summarize, light researchPlusGPT-5.6 Sol is strong; save $80/month
Use Astra’s computer use for real workPro $100Only open Astra tier for new users; 5x Plus usage
Run long coding sessions in CodexPro $100 or a Work planWork plans include Codex allowances; see OpenAI’s Work help article
Burn 20x usage on agentic loopsAPI or wait for Pro $200API: $10/$50 per MTok, cached input $1; no allowance cap, just a bill
Try Astra before committingPlus firstLimited Astra access lets you test quality before upgrading
Already on Pro $200StayCancelling means you cannot re-subscribe until the pause lifts

API as the release valve

Because ChatGPT plans are flat-rate, the pause is really rationing. The API is not paused: GPT-6 Astra costs $10 input / $50 output per million tokens, $1 for cached input, with batch and flex at 50% off, a 1,050,000-token context and 128K max output. Prompts above 272,000 tokens reprice higher. For a developer who would have used Pro $200 for agentic runs, a month of API usage often costs less than $200 — and it scales beyond 20x. Comparison table: GPT-6 Astra vs Claude Fable 5.1 vs Gemini 3.8 Flash pricing.

Alternatives while the queue lasts

  • Claude Max (Anthropic) includes Claude Fable 5.1, GA since September 3, 2026 — the current leader on Artificial Analysis’s Intelligence Index (65.6 vs Astra’s 61.1 at max effort).
  • Google AI Ultra for Gemini 3.8 Flash and 3.5 Pro preview at lower per-task cost.
  • Codex CLI / Claude Code / OpenHands for coding agents that bill through the API rather than a consumer plan. See OpenHands vs Claude Code vs Codex CLI (2026).

What to watch

  1. OpenAI’s help article changing “paused” to open — the only official signal.
  2. Whether OpenAI adds a mid tier (a “10x” plan) or raises Pro $200’s price when it reopens.
  3. New capacity: the first NVIDIA and AMD gigawatt tranches were due in H2 2026.

Sources