Independent model intelligence

Find the right AI model,
not just the loudest one.

Compare current model capability, context, modalities, public preference and API economics in one continuously refreshed decision surface.

LiveBench 2.0 LMArena models.dev OpenRouter AIHOT cross-check

Back to leaderboard

OpenAI · Active

GPT-6 Astra

gpt-astra family · launched 2026-09-04

MultimodalReasoningToolsStructured

AITopic score80.4#5 overall

GPT-6 Astra is OpenAI's most capable model for complex reasoning, coding, computer use, research, and document creation.

Evidence coverage100%9 evaluations · 3 institutions

Context / maximum output1.1M / 128ktokens

Cache / input / output$1 · $10 · $50USD per million tokens

Value / freshness50.8 / 84relative value · release freshness

Product profile

What this model supports

Inputtext, image, pdf

Outputtext

Knowledge cutoff2026-04-30

WeightsProprietary / hosted

Capability profile

Strengths by domain

Coding82.12 public results

Research & reasoning84.72 public results

Professional work75.42 public results

Knowledge & instruction79.82 public results

Behind the score

Every result has a source

Raw results and standardized readings are shown together. Missing tests add neither points nor penalties.

Coding2 results

LiveBench 2.0 · CodingLiveBench93.70 scoreReading 93.7

Imported from LiveBench 2.0. The published 0–100 score is used directly; it is not z-score normalized.

Configuration: LiveBench 2.0, AIHOT aggregateOpen evidence

AIHOT · CodingAIHOT70.40 scoreReading 70.4

Imported from AIHOT aggregate leaderboard. The published 0–100 score is used directly; it is not z-score normalized.

Configuration: LiveBench 2.0, AIHOT aggregateOpen evidence

Research & reasoning2 results

LiveBench 2.0 · ResearchLiveBench92.70 scoreReading 92.7

Imported from LiveBench 2.0. The published 0–100 score is used directly; it is not z-score normalized.

Configuration: LiveBench 2.0, AIHOT aggregateOpen evidence

AIHOT · ResearchAIHOT76.60 scoreReading 76.6

Imported from AIHOT aggregate leaderboard. The published 0–100 score is used directly; it is not z-score normalized.

Configuration: LiveBench 2.0, AIHOT aggregateOpen evidence

Professional work2 results

LiveBench 2.0 · ProfessionalLiveBench83.45 scoreReading 83.5

Imported from LiveBench 2.0. The published 0–100 score is used directly; it is not z-score normalized. Professional = long-agentic, browser, data-analysis and design average.

Configuration: LiveBench 2.0, AIHOT aggregateOpen evidence

AIHOT · ProfessionalAIHOT67.40 scoreReading 67.4

Imported from AIHOT aggregate leaderboard. The published 0–100 score is used directly; it is not z-score normalized.

Configuration: LiveBench 2.0, AIHOT aggregateOpen evidence

Knowledge & instruction2 results

LiveBench 2.0 · KnowledgeLiveBench85.20 scoreReading 85.2

Imported from LiveBench 2.0. The published 0–100 score is used directly; it is not z-score normalized. Knowledge uses the published LiveBench research suite.

Configuration: LiveBench 2.0, AIHOT aggregateOpen evidence

AIHOT · KnowledgeAIHOT74.40 scoreReading 74.4

Imported from AIHOT aggregate leaderboard. The published 0–100 score is used directly; it is not z-score normalized.

Configuration: LiveBench 2.0, AIHOT aggregateOpen evidence

Human preference & indices4 results

LiveBenchLiveBench86.49 scoreReading 86.5

Imported from LiveBench 2.0. The published 0–100 score is used directly; it is not z-score normalized.

Configuration: LiveBench 2.0, AIHOT aggregateOpen evidence

Arena TextLMArena1,441.2 EloReading 80.4

Direct LMArena text leaderboard result.

AIHOT aggregate leaderboardAIHOT71.60 scoreReading 71.6

Imported from AIHOT aggregate leaderboard. The published 0–100 score is used directly; it is not z-score normalized.

Configuration: LiveBench 2.0, AIHOT aggregateOpen evidence

AA IndexArtificial Analysis63.84 indexReading 54.8

Imported from Artificial Analysis. The published 0–100 score is used directly; it is not z-score normalized.

Configuration: LiveBench 2.0, AIHOT aggregateOpen evidence