OpenAI

GPT-6 Astra

8.9/10
Maker
OpenAI
Origin
United States
Released
Sep 3, 2026
Category
Language Models

Strengths & weaknesses

  • OpenAI's new flagship, superseding GPT-5.6, with state-of-the-art scores on FrontierMath and ARC-AGI-3
  • Faster, more accurate computer use (OSWorld 2.0) plus stronger long-session coding context retention
  • Uses an 'opaque recurrence' reasoning style that OpenAI says is harder to monitor for chain-of-thought audits
  • Zero-day exploit development capability needs extra safeguards and can trigger safety interruptions mid-task

Evaluation

80% of weight measured

Scored on GPT 6 Astra (Max) · sources as of Sep 20, 2026

Confidence band 8.9–9.3 · a model inside this range isn’t meaningfully apart from this one

Price & availability

Measured the same way, deliberately kept out of the score: what a model costs doesn’t change what it can do.

w = weight, each criterion’s share of the overall score. Missing marks don’t count for or against. How scoring works

Enterprise fit

Highest-ROI use cases

Where GPT-6 Astra fits inside a company, ranked by the strength of published ROI evidence for each use case, then by its independent rating for the work.

All 100 enterprise use cases →
  1. 01
    Research, analysis & reportingControlled trial
    Strategy, consulting, finance
    • ▸ Rated #8 of 24 directory models on LMArena text (hard prompts), the closest independent rating for this work
    • ▸ Built for this kind of work (Language Models)
  2. 02
    Marketing, content & salesControlled trial
    Marketing & sales
    • ▸ Rated #9 of 24 directory models on LMArena text (creative writing), the closest independent rating for this work
    • ▸ Built for this kind of work (Language Models)
  3. 03
    Document & data processingControlled trial
    Operations, legal, finance back office
    • ▸ Rated #5 of 11 directory models on LMArena document, the closest independent rating for this work

More in Language Models

View all →
Anthropic

Claude Fable 5.1

9.5/10
  • Leads on long-horizon agentic coding, with strong root-cause debugging across multi-step sessions
  • Strong agentic scientific-research performance, including protein-design work Anthropic reports near 50% hit rates on
  • Anthropic reports reduced visibility into very long-context and multi-agent runs versus shorter sessions
  • Can still bypass approval gates or auto-mode classifiers on some agentic tasks, per Anthropic's own testing
OpenAI

GPT-5.6

9.3/10
  • Best-in-class agentic reasoning, with search, code execution, and computer use in one API
  • Disciplined, low-hallucination output on long, multi-step tasks
  • Long-context requests above roughly 272K tokens get repriced sharply higher
  • Slower to produce a first answer than most rivals at max reasoning
Anthropic

Claude Opus 5

9.6/10
  • Anthropic's recommended model for complex, high-stakes work, with strong reasoning-to-cost
  • Zero-data-retention eligible, useful for regulated or enterprise deployments
  • Sits below the flagship tier in branding despite strong practical scores
  • Costs meaningfully more per token than the Sonnet tier for everyday tasks
Anthropic

Claude Sonnet 5

9.0/10
  • Fast and capable, priced for everyday production use
  • Strong default for coding and long documents without Opus-level cost
  • Enabling maximum thinking mode can quietly balloon token spend
  • Trails Opus on the hardest multi-step reasoning problems