// COMPARISON

Qwen3-235B-A22B-Thinking (Jul 2025)

A head-to-head benchmark comparison of Qwen3-235B-A22B-Thinking (Jul 2025) across 31 evaluations.

The models

Benchmark comparison

Benchmark Qwen3-235B-A22B-Thinking (Jul 2025)
code
Codeforces Elo 2,055
LiveCodeBench 78.2%
LiveCodeBench v5 70.7%
LiveCodeBench v6 75.72%
SciCode 42.9%
coding
Aider Benchmark 88.91
general
ArenaHard v2 80.18
C-Eval 93.13%
IFEval 87.8%
MMLU-Pro 84.5%
MMLU-Redux 92.7%
MultiChallenge 52.75%
knowledge
AGIEval 90.01
CMMLU 90.64
TriviaQA 79.63
math
AIME 2024 94.1%
AIME 2025 92.3%
CNMO 2024 89.5%
HMMT February 2025 83.9%
MATH-500 98%
Omni-MATH 82.52
UGMathBench 77
reasoning
ARC-AGI-1 48.12
BBEH 60
FinanceReasoning 87.65
GPQA Diamond 81.1%
Humanity's Last Exam 15.8%
ZebraLogic 97.03%
safety
HealthBench 55.56%
science
PhyBench 42.61
tool-use
BFCL v3 73.53%

Best result per row highlighted in cyan. Each benchmark links to its definition and sources; each model links to its full scorecard.

Frequently asked questions

Which of these models is best for coding? ▶

Qwen3-235B-A22B-Thinking (Jul 2025) has the strongest result on Codeforces Elo among the models compared here. See the code-category rows in the table for the full picture.

How many benchmarks are compared? ▶

This comparison covers 31 benchmarks on which at least one of the selected models has a published score.

Where do the benchmark scores come from? ▶

Scores are aggregated from official model cards, technical reports and standard public evaluations, and link back to each benchmark's source. They are updated as new results are published.

Explore further

→ Qwen3-235B-A22B-Thinking (Jul 2025) → AI benchmarks leaderboard → Build a different comparison
Build a different comparison

Models (0 selected)

Benchmarks (auto)

Select at least two models.