GPT-6 Astra
Made by OpenAI, released 3 Sep 2026. Epoch AI lists its access as “API access”. Appears strongest at knowledge and maths among the models listed here.
In short
What it does well
- Knowledge: Top tier (2nd of 43 listed models)
- Maths: Top tier (1st of 43 listed models)
- Coding: Strong (33rd of 77 listed models)
Where it falls short
- Writing: Middle (57th of 77 listed models)
What we don't know yet
- Fixing real software bugs: Not enough data
The four ratings
- Top tier
- Strong
- Middle
- Behind the leaders
- Not enough data
Ratings compare only the 77 models listed here, in quarters, best first. “Not enough data” means no result yet. It is never a low score.
- Writing
- MiddleHow often people preferred its answers to writing prompts. 1,420 on Arena's scale · Source for GPT-6 Astra writing score (lmarena)
- Coding
- StrongHow often people preferred its answers to coding questions. 1,488 on Arena's scale · Source for GPT-6 Astra coding score (lmarena)
- Maths
- Top tierHow many very hard maths problems it solved. 93.7% · Source for GPT-6 Astra maths score (epoch)
- Knowledge
- Top tierHow many short factual questions it answered correctly. 75.6% · Source for GPT-6 Astra knowledge score (epoch)
Evidence completeness: Broad evidence, 12 of 13 tests have a result.
Where it is relatively stronger or weaker
This compares each kind of request with the model's own overall position among the 77 listed models, using Arena's votes. It is about performance on different languages, kinds of request and topics, not about political or social bias.
Relatively stronger
- Expert-level prompts: 30th of 77 on prompts that need expert knowledge, against 62nd of 77 overall. Source for Expert-level prompts
- Long prompts: 41st of 77 on long, detailed prompts, against 62nd of 77 overall. Source for Long prompts
- Hard prompts: 43rd of 77 on prompts Arena classes as hard, against 62nd of 77 overall. Source for Hard prompts
Relatively weaker
No category stands out as clearly weaker than its overall position.
Show every category we can see
| Category | Position | Arena score | Votes | Source |
|---|---|---|---|---|
| Expert-level prompts Kind of request | 30th of 77 | 1,493 95% range 1,478 to 1,509 | 1,451 | Source for Expert-level prompts |
| Hard prompts Kind of request | 43rd of 77 | 1,462 95% range 1,455 to 1,469 | 9,105 | Source for Hard prompts |
| Long prompts Kind of request | 41st of 77 | 1,461 95% range 1,453 to 1,469 | 6,225 | Source for Long prompts |
| Longer conversations Kind of request | 54th of 77 | 1,448 95% range 1,435 to 1,462 | 2,103 | Source for Longer conversations |
| Chinese Language | 50th of 76 | 1,492 95% range 1,475 to 1,509 | 1,185 | Source for Chinese |
| English prompts Language | 63rd of 77 | 1,447 95% range 1,439 to 1,455 | 6,008 | Source for English prompts |
| French Language | 63rd of 72 | 1,435 95% range 1,407 to 1,463 | 473 | Source for French |
| Non-English prompts Language | 54th of 77 | 1,432 95% range 1,425 to 1,439 | 9,026 | Source for Non-English prompts |
| Russian Language | 56th of 77 | 1,441 95% range 1,427 to 1,455 | 1,938 | Source for Russian |
| Spanish Language | 63rd of 70 | 1,419 95% range 1,392 to 1,445 | 479 | Source for Spanish |
| Entertainment, sport and media Topic | 44th of 77 | 1,427 95% range 1,417 to 1,437 | 3,837 | Source for Entertainment, sport and media |
| Law and government Topic | 47th of 76 | 1,455 95% range 1,438 to 1,472 | 1,263 | Source for Law and government |
| Medicine and health Topic | 66th of 75 | 1,419 95% range 1,400 to 1,438 | 1,067 | Source for Medicine and health |
| Science Topic | 63rd of 77 | 1,450 95% range 1,437 to 1,462 | 2,543 | Source for Science |
| Software and IT Topic | 44th of 77 | 1,475 95% range 1,466 to 1,483 | 5,349 | Source for Software and IT |
Bias: what we can and can't see
Our sources do not measure political lean, stereotyping, flattery (agreeing with you too readily) or safety behaviour, so we cannot say how any model does on those. What we can show is above: where a model is relatively stronger or weaker across languages, kinds of request and topics. Also, people voting in Arena can favour answer style, such as length and formatting, and our numbers do not adjust for that. See what AI bias means and what these tests can tell you.
Best-supported use cases
- Homework and maths: good-enough alternative.
Where an alternative might be more suitable
- Writing and creating content: rated Middle on writing votes. Compare it with Claude Opus 5.5, the best-supported choice for this task.
- Coding and building things: rated Strong on coding votes. Compare it with Claude Opus 5.5, the best-supported choice for this task.
- Business emails and everyday work: rated Middle on business-task votes. Compare it with Gemini 4 Argon, the best-supported choice for this task.
What it is
- Maker
- OpenAI
- Release date (Epoch AI)
- 3 Sep 2026
- Access, as listed by Epoch AI
- API access
- Licence, as listed by Arena
- Proprietary
- Maker's country (Epoch AI)
- United States of America
- Last observation in our data
- 28 Oct 2026
- Data last changed
- 29 Oct 2026, 14:36 UTC
We have not verified an official product page for this model, so we do not link to one.
How you can use it
Proprietary. The model's files are not published, so it is generally available only through its maker or partners the maker chooses.
How you can reach it (an app, a website, a developer API or a third-party tool) and what that costs are not covered by our benchmark data. Where we could verify prices from the maker's own pages, they are shown under Price. Epoch AI lists access as “API access”. See models, apps and harnesses and how AI pricing works.
Price
Verified consumer plans that include this model
- ChatGPT Plus: GBP 20.00 per month. Maker's page, checked 29 Oct 2026. Comparison table cell for this model on this plan: "Yes". The page says limits apply.
- ChatGPT Pro (from): GBP 89.00 per month. Maker's page, checked 29 Oct 2026. Comparison table cell for this model on this plan: "Expanded". Listed as 'From £89 / month' with a choice of 3 usage tiers.
Prices change and can differ by country. Check the maker's page before you pay.
See ChatGPT plans and pricing (checked 29 Oct 2026). Page opened and read in the browser. Which plan includes which model comes from its comparison table.
We never infer subscription prices from API prices. New to “tokens” and “per million”? See how AI pricing works.
Good to remember
- Benchmarks are samples of tasks. None proves performance in every situation.
- Arena scores come from people choosing between answers. That measures preference, not correctness.
- Ratings are relative to the 77 models listed here. A different set would change them.
- Sources test models at different settings. Each result shows the setting used, and we do not adjust it.
The evidence, for the curious
Show all 13 tests with settings and sources
| Test | Result | Rating | Setting used | Source identifier | Observed | Links |
|---|---|---|---|---|---|---|
| Arena text: overall preference Arena 'text' leaderboard (Bradley-Terry rating, no style control) |
1,441 Arena score 95% range 1,435 to 1,447, 15,038 votes |
Behind the leaders | effort: Max, bradley-terry rating | GPT-6 Astra (Max) lmarena · text/latest |
28 Oct 2026 | Source for GPT-6 Astra Arena text: overall preference score (lmarena) |
| Arena text: creative writing Arena 'text' leaderboard, category creative_writing |
1,420 Arena score 95% range 1,409 to 1,431, 3,323 votes |
Middle | effort: Max, bradley-terry rating | GPT-6 Astra (Max) lmarena · text/latest |
28 Oct 2026 | Source for GPT-6 Astra Arena text: creative writing score (lmarena) |
| Arena text: coding Arena 'text' leaderboard, category coding |
1,488 Arena score 95% range 1,477 to 1,499, 3,393 votes |
Strong | effort: Max, bradley-terry rating | GPT-6 Astra (Max) lmarena · text/latest |
28 Oct 2026 | Source for GPT-6 Astra Arena text: coding score (lmarena) |
| Arena text: maths Arena 'text' leaderboard, category math |
1,468 Arena score 95% range 1,448 to 1,488, 804 votes |
Middle | effort: Max, bradley-terry rating | GPT-6 Astra (Max) lmarena · text/latest |
28 Oct 2026 | Source for GPT-6 Astra Arena text: maths score (lmarena) |
| Arena text: following instructions Arena 'text' leaderboard, category instruction_following |
1,454 Arena score 95% range 1,445 to 1,463, 4,755 votes |
Strong | effort: Max, bradley-terry rating | GPT-6 Astra (Max) lmarena · text/latest |
28 Oct 2026 | Source for GPT-6 Astra Arena text: following instructions score (lmarena) |
| Arena text: business and management Arena 'text' leaderboard, category industry_business_and_management_and_financial_operations |
1,438 Arena score 95% range 1,426 to 1,450, 2,613 votes |
Middle | effort: Max, bradley-terry rating | GPT-6 Astra (Max) lmarena · text/latest |
28 Oct 2026 | Source for GPT-6 Astra Arena text: business and management score (lmarena) |
| Arena WebDev: building web apps Arena 'webdev' leaderboard (Code Arena), category overall |
1,773 Arena score 95% range 1,764 to 1,783, 7,210 votes |
Top tier | effort: Max, bradley-terry rating | GPT-6 Astra (Max) lmarena · webdev/latest |
28 Oct 2026 | Source for GPT-6 Astra Arena WebDev: building web apps score (lmarena) |
| FrontierMath Tiers 1-3 (v2) FrontierMath Tiers 1-3 v2 (private set), benchmark released 2026-06-12 |
93.7% share of problems solved standard error ±1.4 points |
Top tier | effort: max, best score across scorers | gpt-6-astra_max epoch · frontiermath_tiers_1_3_v2.csv |
30 Aug 2026 | Source for GPT-6 Astra FrontierMath Tiers 1-3 (v2) score (epoch) No public run log in the data file. |
| OTIS Mock AIME 2024-2025 OTIS Mock AIME 2024-2025, benchmark released 2024-12-19 |
100.0% share of problems solved standard error ±0.0 points |
Top tier | effort: max, best score across scorers | gpt-6-astra_max epoch · otis_mock_aime_2024_2025.csv |
30 Aug 2026 | Source for GPT-6 Astra OTIS Mock AIME 2024-2025 score (epoch) No public run log in the data file. |
| GPQA Diamond GPQA Diamond, benchmark released 2023-11-20 |
95.8% share of questions correct standard error ±1.4 points |
Top tier | effort: max, best score across scorers | gpt-6-astra_max epoch · gpqa_diamond.csv |
30 Aug 2026 | Source for GPT-6 Astra GPQA Diamond score (epoch) No public run log in the data file. |
| SimpleQA Verified SimpleQA Verified, benchmark released 2025-09-09 |
75.6% share of questions correct standard error ±1.4 points |
Top tier | effort: max, best score across scorers | gpt-6-astra_max epoch · simpleqa_verified.csv |
30 Aug 2026 | Source for GPT-6 Astra SimpleQA Verified score (epoch) No public run log in the data file. |
| SWE-bench Verified (Epoch runs) | Not enough data | |||||
| FrontierCode (reported by Cognition) FrontierCode, benchmark released 2026-06-08, aggregation Mean@5 |
53.3% mean score over 5 attempts no uncertainty figure published |
Not rated (results use different tools) | effort: max, tool: codex, mean@5 | gpt-6-astra_max epoch · frontiercode_external.csv |
not provided | Source for GPT-6 Astra FrontierCode (reported by Cognition) score (epoch) · Run details for FrontierCode (reported by Cognition) |
Missing evidence
Not enough data for these tests:
- SWE-bench Verified (Epoch runs)
Missing results are not low scores and do not count against the model.
Where a source gives no public run log, we cannot show the run behind the score. See data sources.
Not sure this is the right one?
Tell us what you need help with and we'll narrow the options down.