GPT-6 Astra
GPT-6 Astra is OpenAI's most capable model for complex reasoning, coding, computer use, research, and document creation.
- Strongest at
- Reasoning Top 2% of the past year's models · 4 of 5 benchmarks
- Price per 1M tokens
- $10 / $50 input / output · OpenAI
- Context window
- 1.05M tokens in · 128K out
Capability profile
Scored against 91 models released in the past year. 50 is the average model; further out is better.
Where these scores come from
Each capability combines the published benchmarks listed here, rescaled so 50 is the reference average and 15 points is one standard deviation. A hollow dot rests on a single benchmark.
- Coding DeepSWE, FrontierCode, GSO-Bench, MirrorCode, LiveBench: Coding + Agentic Coding
- Writing EQ-Bench Creative Writing v3, EQ-Bench Longform Writing, Lech Mazur Writing comparison, LiveBench: Instruction Following
- Reading & documents GDP.pdf (Surge AI), Vision Evals: OCR & extraction (low effort), LiveBench: Language + Data Analysis, ExtractBench: document extraction
- Reasoning ARC-AGI-2, Chess Puzzles, Mystery Game Puzzles, LiveBench: Reasoning
- Math OTIS Mock AIME 2024-2025, FrontierMath-Tiers-1-3-v2-Private, FrontierMath-Tier-4-v2-Private, ProofBench, LiveBench: Mathematics
- Science GPQA diamond, HLE, Terminal-Bench Science 0.1, WeirdML v2
- Visual reasoning Vision Evals (low effort), Blueprint-Bench 2, GDP.pdf (Surge AI)
- Computer use Terminal-Bench 4.0, APEX-Agents, AutomationBench 1.0.6
Price against capability
The shaded corner holds anything better and cheaper. Prices blend input and output 3 to 1.
1 model from the past year scores higher on Reasoning for less: GPT-6.1 Terra.
1 model from the past year scores higher on Coding for less: Claude Opus 5.5.
1 model from the past year scores higher on Writing for less: Kimi K3.1 Pro.
1 model from the past year scores higher on Reading & documents for less: GPT-6.1 Sol.
1 model from the past year scores higher on Math for less: GPT-6.1 Sol.
1 model from the past year scores higher on Science for less: Claude Opus 5.5.
2 models from the past year score higher on Visual reasoning for less: GPT-6.1 Terra, GPT-6.1 Sol.
4 models from the past year score higher on Computer use for less: Gemini 4 Argon, Qwen3.8 Max, Claude Opus 5.5 and 1 more.
OpenAI $10 / $50 · OpenRouter $10 / $50 · Cached input $1
The GPT-6 Astra line
2 releases in the GPT-6 Astra line.
A newer release is out: GPT-6 Astra (Fast).
Epoch benchmark boards
Best result on each board it has run. The tick marks the leader.
15 more boards
OpenAI's past year
24 release days in the past year, about one every 15 days. The next is expected around Nov 8, 2026. Forecasts ↗
More from the lab
Related article · Sep 9, 2026GPT-6 Astra: The next generation in intelligence for workSpecifications
Specifications and prices from Models.dev; availability and limits may differ by provider. Benchmarks from Epoch AI (CC BY 4.0), LiveBench and the other sources named above.