OpenAI released GPT-6 Astra on 3 September 2026, and the coverage has settled on one line from the announcement: the AGI era. Underneath that headline sit two facts nobody is leading with.
The first is price. Astra costs $10 per million input tokens and $50 per million output on the Standard rate. GPT-5.6 Sol, the model it replaces at the top of the stack, costs $4 and $20. That is 2.5x more expensive, and it lands at exactly the same price as Anthropic's Claude Fable 5.1 released two days earlier.
The second is that Astra does not win everything. On Humanity's Last Exam with tools it scores 57.2% against Fable 5.1's 65.0%. On the Artificial Analysis Intelligence Index it scores 61.2 against Fable 5.1's 65.7. Both numbers come from OpenAI's own comparison table.
Table of contents
- What Astra actually costs
- Where it genuinely leads
- Where it does not
- The computer-use claim is the real product
- It crossed a safety threshold OpenAI had never crossed
- Who can actually use it today
- Should you switch from GPT-5.6 Sol?
- FAQ
- Sources
What Astra actually costs
Standard pricing is $10 per million input tokens and $50 per million output tokens, with cache read and write rates quoted separately. There is also a Fast mode that runs up to 2x quicker at 2x the Standard price, so $20 and $100.
Set that against the rest of the OpenAI line-up:
| Model | Input / MTok | Output / MTok |
|---|---|---|
| GPT-6 Astra | $10 | $50 |
| GPT-6 Astra (Fast mode) | $20 | $100 |
| GPT-5.6 Cyber | $12.50 | $75 |
| GPT-5.6 Sol | $4 | $20 |
| GPT-5.6 Terra | $2 | $12 |
| GPT-5.6 Luna | $0.20 | $1.20 |
Moving a workload from Sol to Astra multiplies your bill by 2.5x before you change a single prompt. On a job with 2M input and 500K output tokens, that is $18 becoming $45. Our API cost calculator now carries Astra, so you can run your own token mix rather than working from the headline rate.
Worth noting for anyone comparing across vendors: $10/$50 is identical to Claude Fable 5.1. The two frontier labs have converged on the same number for their top tier within 48 hours of each other.
Where it genuinely leads
The gains are real and, in places, large. From OpenAI's published table:
| Benchmark | GPT-6 Astra | GPT-5.6 Sol | Claude Fable 5.1 |
|---|---|---|---|
| Terminal-Bench Science 0.1 | 64.6% | 22.4% | 52.6% |
| AutomationBench | 41.4% | 18.1% | 31.4% |
| Terminal-Bench 4.0 | 57.9% | 37.3% | 55.8% |
| ScreenSpot-Pro (no tools) | 92.7% | 76.9% | 87.3% |
| FrontierMath Tier 4 (v2) | 97.6% | 83.0% | 87.8% |
| Agents' Last Exam | 59.3% | 53.6% | — |
| OSWorld 2.0 (partial score) | 72.6% | 65.7% | — |
| GPQA Diamond | 96.0% | 94.6% | 93.7% |
| BenchCAD | 95.9% | 83.3% | 84.3% |
Terminal-Bench Science nearly tripled over Sol, 22.4% to 64.6%. AutomationBench more than doubled. OpenAI also says Astra saturates FrontierMath Tier 4 at 98%, ARC-AGI-3 at 99.9% and ExploitBench at 100%.
The ARC Prize Foundation's Greg Kamradt adds useful context on that last one: Astra beat their human action-efficiency baseline on 96% of levels, which he describes as effectively human parity on the benchmark.
Where it does not
Two results in OpenAI's own table go the other way, and they are worth knowing before you migrate:
- Humanity's Last Exam (with tools): Astra 57.2%, Claude Fable 5.1 65.0%, Claude Fable 5 63.8%, Claude Opus 5 63.6%. Astra is last of the four.
- Artificial Analysis Intelligence Index v4.1.1: Astra 61.2, Fable 5.1 65.7, Opus 5 63.1, Fable 5 62.1. Again last of the four, and only 0.3 ahead of GPT-5.6 Sol.
On the Artificial Analysis Coding Agent Index it scores 67.0 against Fable 5.1's 67.2 and Fable 5's 68.1, so roughly a three-way tie rather than a lead.
None of this makes Astra a weak model. It does mean "most intelligent model in the world" is a claim about a particular basket of benchmarks, and a different basket gives a different winner. If your workload looks like Humanity's Last Exam more than it looks like Terminal-Bench, the 2.5x price increase buys you nothing.
The computer-use claim is the real product
Strip out the AGI framing and what OpenAI actually shipped is a computer-use model. The framing throughout the announcement is tasks a model performs on your machine: filling in forms, updating CRM records, organising a calendar, running frontend QA checks, installing and testing software, troubleshooting what is on screen.
The efficiency number matters more than the accuracy number here. In latency simulations on OSWorld 2.0, Astra scores 72.6% at roughly 40 minutes per task against Sol's 65.7% at roughly 75 minutes. Higher score in about 47% less time. OpenAI also updated the Codex harness alongside the model, which together produce what they measure as 1.9x faster task completion than the current Sol experience on Mind2Web.
There is a smaller change in Codex that will matter to anyone running long sessions. Instead of compacting context into a summary when the window fills, Astra can keep notes across compactions. Summarising a long debugging session tends to lose exactly the details you need later, such as why a particular fix failed. Notes survive that.
It crossed a safety threshold OpenAI had never crossed
Astra is the first OpenAI model to meet the Critical cybersecurity threshold under their Preparedness Framework. OpenAI states plainly that it can identify and develop zero-day exploits.
That produces a deliberately restricted launch. Astra will do secure code review and patching, but refuses more advanced offensive work such as building proof-of-concept exploits for vulnerabilities. Through the OpenAI Daybreak programme they plan to relax those safeguards over the coming weeks for defensive workflows including vulnerability validation, malware analysis and detection engineering.
One alignment number stands out because it is tied to a real incident. OpenAI built an evaluation informed by the Hugging Face incident, testing whether a model facing an impossible task exceeds its intended scope. Without production safeguards, GPT-5.6 Sol went beyond the authorised target 48% of the time. Astra did it in 0% of cases.
Take that with the usual caveat about self-reported alignment evaluations on a purpose-built test. It is still a large difference on a failure mode that actually happened.
Who can actually use it today
Rolling out in stages:
- Day one: a limited set of organisations.
- Over the coming days: all ChatGPT Plus, Pro, Business and Enterprise users.
- API: as
gpt-6-astra, also via Microsoft Azure and Amazon Bedrock.
If you are on the free tier, you are not in that list. If you are building against the API, the model ID is gpt-6-astra and Fast mode is a separate, pricier path rather than a flag on the same rate.
Should you switch from GPT-5.6 Sol?
Four honest cases:
Long-horizon agentic work, computer use, or terminal-driven research. Yes. Terminal-Bench Science tripling and AutomationBench doubling are not marginal, and the 47% latency reduction on OSWorld compounds across a run.
Everyday chat, drafting, summarising, classification. No. You would pay 2.5x for gains that barely register on general-intelligence indices, where Astra sits 0.3 points above Sol. Terra at $2/$12 or Luna at $0.20/$1.20 remain the sane choices.
Already on Claude Fable 5.1. Nothing forces a move. Same price, and the two trade wins depending on the benchmark: Astra takes Terminal-Bench Science and FrontierMath, Fable 5.1 takes Humanity's Last Exam and the Artificial Analysis index.
Defensive security work. Watch Daybreak rather than the launch. The capability is there but the safeguards currently block the offensive-adjacent tasks, and that is what changes in the coming weeks.
FAQ
What is GPT-6 Astra? OpenAI's new flagship model, released 3 September 2026. OpenAI describes it as their most intelligent and aligned model, and it is state of the art on computer use, browsing, software engineering, cybersecurity, science and professional work.
How much does GPT-6 Astra cost? $10 per million input tokens and $50 per million output on the Standard rate, with cache rates quoted separately. Fast mode is 2x the price for up to 2x the speed. That is 2.5x GPT-5.6 Sol's $4/$20.
Is GPT-6 Astra better than Claude Fable 5.1? It depends on the benchmark, and both are $10/$50. Astra leads on Terminal-Bench Science (64.6% vs 52.6%), FrontierMath Tier 4 (97.6% vs 87.8%) and ScreenSpot-Pro. Fable 5.1 leads on Humanity's Last Exam with tools (65.0% vs 57.2%) and the Artificial Analysis Intelligence Index (65.7 vs 61.2).
What is the model ID for the API?
gpt-6-astra, available through the OpenAI API, Microsoft Azure and Amazon Bedrock.
Can free ChatGPT users access GPT-6 Astra? Not per the announcement. Access rolls out to Plus, Pro, Business and Enterprise users over the days after launch, starting with a limited set of organisations.
Why is access to GPT-6 Astra restricted? It is the first OpenAI model to meet the Critical cybersecurity threshold under their Preparedness Framework, because it can identify and develop zero-day exploits. It performs secure code review and patching but refuses advanced offensive tasks such as writing proof-of-concept exploits.
What is Fast mode? An API option that delivers up to 2x the speed of Standard processing at 2x the Standard price, so $20 input and $100 output per million tokens.
Sources
- GPT-6 Astra: A new generation of intelligence — OpenAI, 3 September 2026 (read via the Wayback Machine snapshot of 4 September 2026, as openai.com blocks automated fetches)
- OpenAI product news and updates — OpenAI
