Benchmark Almanac

Qwen3-VL-235B-A22B-Instruct

Qwen3-VL-235B-A22B-Instruct ranks first on CompassArena Multimodal.

Qwen Released September 23rd, 2025 Family Qwen3-VL The lab's model card Compare with others
22
Benchmarks with a score
1
Ranked first, out of five or more models
2
In the top three
Qwen Sep 2025

Qwen3-VL-235B-A22B-Instruct

Open weights Flagship

No Epoch Capabilities Index score yet
  • MMLongBench-Doc 57% #6 /17
  • RefAdv-S 59.3% #8 /19
  • ClockBench 39.4% #13 /37

Where it ranks

Its best score on each benchmark, and where that ranks among every model with a score there. We rank each model once, at its best setting.

Knowledge and language

BenchmarkScoreRankSource
LMArena (Elo) 1,413 #116 of 335 Official leaderboard
SEA-HELM 71.0 #32 of 71 Official leaderboard
QIMMA 52.9% #12 of 25 Official leaderboard

Agents and tools

BenchmarkScoreRankSource
SnakeBench 29 #48 of 188 Official leaderboard
GameHorizon 63.4 #26 of 37 Official leaderboard
ADeptS-Bench Desktop 37.2% #6 of 7 Official leaderboard
ADeptS-Bench Mobile 34.7% #6 of 7 Official leaderboard

Professional work

BenchmarkScoreRankSource
CADGenBench 16.8% #11 of 11 Official leaderboard

Vision, audio and robotics

BenchmarkScoreRankSource
Math-VR 65% #2 of 26 Official leaderboard
CompassArena Multimodal 1,173 #1 of 8 Official leaderboard
SpaCE-10 47.9% #9 of 36 Official leaderboard
WorldVQA 23.5% #6 of 22 Official leaderboard
MATH-Vision (MathVision) 66% #29 of 105 Official leaderboard
ClockBench 39.4% #13 of 37 Official leaderboard
MMLongBench-Doc 57% #6 of 17 Official leaderboard
RefAdv-S 59.3% #8 of 19 Official leaderboard
Vision Arena 1,214 #70 of 136 Official leaderboard
OmniDocBench v1.5 89.2 #6 of 11 Official leaderboard
OmniDocBench v1.6 89.8 #5 of 8 Official leaderboard
DISBench 11.5% #6 of 9 Official leaderboard
Math Kangaroo 2025 (MathArena) 72.5% #17 of 22 Official leaderboard

Safety and security

BenchmarkScoreRankSource
nocot-bench (NCRI) 77.3 #67 of 185 Official leaderboard