GPT-6 Astra
OpenAI · released Sep 4, 2026
GPT-6 Astra is OpenAI's flagship model for demanding end-to-end work. It is suited for advanced analysis, software engineering, deep research, scientific work, and document creation, with particular strengths in long-horizon...
Specification
- Context window
- 1.05M
- Max output
- 128K
- Knowledge cutoff
- Not stated
- Parameters
- Undisclosed
- Licence
- Proprietary
- Serving providers
- 3
- Moderated
- Yes
- Uptime
- 100.0%
Intelligence
52.7
99th percentile
Coding
76.9
Coding Index
Agentic
51.0
Agentic Index
Output speed
150 t/s
Median across providers
Latency
2.08s
Time to first token
Cost per task
$4.50
Estimated
Benchmarks
Where the score comes from
The Intelligence Index is a composite. These are the underlying evaluations this model was actually measured on.
Evaluation scores
Percentage correct · higher is better
- GPQA Diamond96.1%
- AA-LCR (long context)80.7%
- SciCode56.5%
- Humanity's Last Exam54.7%
- τ²-bench (Telecom)41.4%
An evaluation missing from this list was not run for this model — it is not a zero.
View as table
| Evaluation | Score |
|---|---|
| GPQA Diamond | 96.1% |
| AA-LCR (long context) | 80.7% |
| SciCode | 56.5% |
| Humanity's Last Exam | 54.7% |
| τ²-bench (Telecom) | 41.4% |
Against its peers
Intelligence Index · this model highlighted, nearest peers in grey
- Claude Sonnet 5.5Anthropic56.0
- Claude Fable 5.1Anthropic53.4
- GPT-6 AstraOpenAI52.7
- Gemini 4 ArgonGoogle52.6
- Muse WonderMeta52.4
- GPT-6.1 SolOpenAI51.8
- Claude Opus 5Anthropic50.8
- Claude Fable 5Anthropic49.6
Peers are the models sitting closest on the Intelligence Index — the set you would realistically choose between.
View as table
| Model | Intelligence |
|---|---|
| Claude Sonnet 5.5 | 56.0 |
| Claude Fable 5.1 | 53.4 |
| GPT-6 Astra | 52.7 |
| Gemini 4 Argon | 52.6 |
| Muse Wonder | 52.4 |
| GPT-6.1 Sol | 51.8 |
| Claude Opus 5 | 50.8 |
| Claude Fable 5 | 49.6 |
Percentile among all indexed models
Pricing
What it costs to run
List prices per million tokens, plus what one representative task works out to.
List price
- Input / 1M tokens
- $10
- Output / 1M tokens
- $50
- Cached input / 1M
- $1
- Blended 3:1
- $20
One task, estimated
$4.50
- Input tokens
- 50,000
- Output tokens
- 80,000
- Profile
- Reasoning
Estimated from list pricing: 50K input tokens plus 80K output tokens for reasoning models (25K for non-reasoning).
Serving providers
Speed and latency figures are medians across these providers, so a widely-served model reports a blend rather than any single endpoint.
Arena
Head-to-head generation quality
Elo from pairwise judgements, broken out by the kind of thing the model was asked to build.
- Overall Elo
- 1,468
- Win rate
- 75.3%
- Strongest at
- 3D scenes
- Tournaments
- 1,661
Elo by category
Dot position on a 1,300–1,475 scale · Elo has no meaningful zero
- 1,468
- 1,450
- 1,430
- 1,396
- 1,334
- 1,329
- 1,357
- 1,313
Agent categories (full-stack apps, mobile apps) are only scored for models tested in agent harnesses.
View as table
| Category | Elo | Win rate |
|---|---|---|
| 3D scenes | 1468 | 75.3% |
| Game development | 1450 | 72.7% |
| SVG | 1430 | 68.6% |
| UI components | 1396 | 65.6% |
| Data visualisation | 1334 | 60.4% |
| Websites | 1329 | 57.8% |
| Full-stack apps | 1357 | 61.1% |
| Mobile apps | 1313 | 65.0% |