HolyGrAIL

All tasks / Models

GPT-6 Astra

OpenAI's most capable GPT tier, introduced with GPT-6, aimed at complex reasoning and coding.

  • OpenAI
  • Closed
  • Released 3 Sep 2026

Official page

Where it ranks

Facts

Vendor
OpenAI
Family
GPT Astra
Released
3 Sep 2026
Weights
Closed: API and apps only
License
proprietary
API price
$10 in / $50 out per 1M tokens
Access
Paid only
Last verified
18 Oct 2026

Price from developers.openai.com, 18 Oct 2026.

Benchmark results

BenchmarkResultEntry on the leaderboardAs ofSource
Artificial Analysis Intelligence Index53GPT-6 Astra (Max)
Raw value from the page's embedded model data; the site displays it rounded to an integer.
18 Oct 2026artificialanalysis.ai
Artificial Analysis MMMU-Pro (independent evaluation)86.9%GPT-6 Astra (Max)
Raw fraction 0.8688 from the model data embedded in https://artificialanalysis.ai/leaderboards/models, converted to percent; rank 2 of 124 entries with an MMMU-Pro result; no last-updated date shown.
19 Oct 2026artificialanalysis.ai
AutomationBench-AA (Artificial Analysis)68.5%GPT-6 Astra (Max)
AutomationBench-AA Score (share of each task's objectives completed, a task with any guardrail violation scoring zero); raw value from the page's embedded model data (displayed rounded); rank 7 of 32; Tasks Completed 41.55%, Objectives Completed 88.83%; no last-updated date shown, so as_of = retrieval date. Best of AA's GPT-6 Astra effort variants (Xhigh 67.18, High 66.61).
19 Oct 2026artificialanalysis.ai
Design Arena Python-PPTX Slides1,246GPT-6 Astra (xhigh)
802 battles. Read from the site's own API (page renders client-side).
18 Oct 2026designarena.ai
Design Arena SVG1,431GPT-6 Astra (xhigh)
Design Arena SVG, rank 2, 1621 battles; rows read from POST https://www.designarena.ai/api/leaderboard (arenaType 'models', category 'svg'); names from https://www.designarena.ai/api/registry
19 Oct 2026designarena.ai
EQ-Bench Creative Writing v32,173gpt-6-astra
Elo score from the leaderboard data embedded in https://eqbench.com/creative_writing.js (file last modified 2026-09-24); rank 1 of 140; rubric score 16.80. Marked as a new entry on the board.
24 Sep 2026eqbench.com
LMArena Search Arena (style control)1,246gpt-6-astra-max
From LMArena's official leaderboard-dataset (search_style_control, split latest, category overall, published 2026-10-13), which mirrors https://arena.ai/leaderboard/search (site blocks automated fetches). Rank 3 of 42; 18483 votes; 95% CI 1238.2-1254.3.
13 Oct 2026huggingface.co
LMArena Text Arena (overall, style control)1,477gpt-6-astra-max
From LMArena's official leaderboard-dataset (text_style_control, split latest, category overall, published 2026-10-15), which mirrors https://arena.ai/leaderboard/text (site blocks automated fetches).
15 Oct 2026huggingface.co
LMArena Text Arena: Creative Writing (style control)1,450gpt-6-astra-max
From LMArena's official leaderboard-dataset (text_style_control, split latest, category creative_writing, published 2026-10-15), which mirrors https://arena.ai/leaderboard/text/creative-writing (site blocks automated fetches). Rank 38 of 414; 2718 votes; 95% CI 1437.7-1462.2.
15 Oct 2026huggingface.co
LMArena Vision Arena (overall, style control)1,286gpt-6-astra-max
From LMArena's official leaderboard-dataset (vision_style_control, split latest, category overall, published 2026-10-09), which mirrors https://arena.ai/leaderboard/vision (site blocks automated fetches). Rank 19 of 165; 3590 votes; 95% CI 1275.2-1296.8.
9 Oct 2026huggingface.co
LMArena WebDev / Code Arena (overall)1,776GPT-6 Astra (Max)
From LMArena's official leaderboard-dataset (webdev = Code Arena, split latest, category overall, published 2026-10-18), which mirrors https://arena.ai/leaderboard/webdev. Rank 2; 6172 votes; 95% CI 1765.5-1785.8.
18 Oct 2026huggingface.co
ParseBench (models)70.7%OpenAI GPT-6 Astra (Reasoning Low)
Overall (mean of tables, charts, content faithfulness, semantic formatting, visual grounding) from leaderboard.csv in the ParseBench repo (last leaderboard commit 2026-10-09).
9 Oct 2026github.com
SWE-Bench Pro V2 (public set, full)96.9%GPT-6-Astra (Codex) high
SWE-Bench Pro V2 Full (642 tasks); row added 2026-10-07; 95% CI +/-1.1; harness in the variant name; read from the leaderboard page's embedded data.
19 Oct 2026labs.scale.com

Where you can use it

Compare

Alternatives to GPT-6 Astra

Sources