supermarket

Language

Qwen3-235B-A22B-Thinking-2507

By Alibaba Group, in the Qwen family.

Released
25 Jul 2025
Status
Legacy
Context
262.1K
Max output
—
Input per 1M tokens
$0.23
Output per 1M tokens
$2.30
Open weights
Yes
License
apache-2.0
Parameters
235B, 22B active
Knowledge cutoff
—
Reasoning
Yes
Modalities
text → text

Category scores

Out of 100, from independent and third-party results only, by method category-2026-09-29.1.

Category scores
CategoryScoreRankCoverage
Knowledge & reasoning20.9772 of 4
Factuality & hallucination40.4381 of 1
Human preference50.0Not ranked1 of 5

Benchmark results

Every result we hold for this model, with who ran it and where it was published.

Knowledge & reasoning

Knowledge & reasoning results
BenchmarkMetricValueProvenanceSettingEvaluatedSource
Chess PuzzlesPuzzles solved12.0%Independent—11 Dec 2025Epoch AI
DTBench (Decision Theory Benchmark)Accuracy (capability questions)80.3%Third party——https://conceptualreasoning.ai/dtbench via Epoch AI
LMCA (Language Model Conceptual Argumentation)Correlation with expert ratings (Pearson r)0.2931Third party——https://conceptualreasoning.ai/lmca via Epoch AI
Mystery Game PuzzlesPuzzles solved9.0%Independent—27 Aug 2026Epoch AI

Scientific reasoning

Scientific reasoning results
BenchmarkMetricValueProvenanceSettingEvaluatedSource
GPQA DiamondAccuracy80.1%Independent—11 Dec 2025Epoch AI

Mathematics

Mathematics results
BenchmarkMetricValueProvenanceSettingEvaluatedSource
FrontierMath Tier 4 (v1)Accuracy0.0%Independent—11 Dec 2025Epoch AI
FrontierMath Tiers 1-3 (v1)Accuracy8.5%Independent—11 Dec 2025Epoch AI
OTIS Mock AIME 2024-2025Accuracy86.7%Independent—10 Dec 2025Epoch AI

Coding & software engineering

Coding & software engineering results
BenchmarkMetricValueProvenanceSettingEvaluatedSource
WeirdML (v2)Average accuracy41.0%Third party——WeirdML Leaderboard via Epoch AI
WeirdML (v2)Average accuracy38.7%Third party——WeirdML Leaderboard via Epoch AI
WeirdML (v2)Average accuracy37.3%Third party——WeirdML Leaderboard via Epoch AI

Long context

Long context results
BenchmarkMetricValueProvenanceSettingEvaluatedSource
Fiction.LiveBench% correct at 16K tokens (Epoch's column)75.0%Third party——Fiction.live leaderboard via Epoch AI

Factuality & hallucination

Factuality & hallucination results
BenchmarkMetricValueProvenanceSettingEvaluatedSource
SimpleQA VerifiedAccuracy (% correct)40.4%Independent—27 Aug 2026Epoch AI

Writing

Writing results
BenchmarkMetricValueProvenanceSettingEvaluatedSource
LLM Creative Story-Writing Benchmark V1 (Lech Mazur)Mean rubric score (percent of the 10-point scale)82.4%Third party——lechmazur/writing Github repository via Epoch AI

Human preference

Human preference results
BenchmarkMetricValueProvenanceSettingEvaluatedSource
Text Arena (arena.ai)Arena score (Elo scale)1400Independent—25 Sep 2026Arena
Text Arena (arena.ai)Votes8776Independent—25 Sep 2026Arena

Other Qwen versions

Sources