MMLU-ProX Benchmark

MMLU-ProX Benchmark | Phaseo

Models Using This Benchmark

Organisation	Model	Reported	Top Score	Info	Self Reported	Source
Qwen	Qwen 3.5 397B A17B	16 Feb 2026	84.70%	-	Yes	Source
Qwen	Qwen 3.6 Plus	01 Apr 2026	84.70%	-	Yes	Source
Qwen	Qwen 3.5 122B A10B	24 Feb 2026	82.20%	-	Yes	Source
Qwen	Qwen 3.5 27B	24 Feb 2026	82.20%	-	Yes	Source
Qwen	Qwen 3.5 Flash	23 Feb 2026	82.20%	inferred family alias from qwen3.5-27b (score=0.4147; benches=81)	Yes	Source
Qwen	Qwen 3 235B A22B Thinking 2507	-	81%	MMLU-ProX	Yes	Source
Qwen	Qwen 3.5 35B A3B	24 Feb 2026	81%	-	Yes	Source
Qwen	Qwen 3 VL 235B A22B Thinking	-	80.60%	-	Yes	-
Nvidia	Nemotron 3 Super	11 Mar 2026	79.36%	-	Yes	Source
Qwen	Qwen 3 Next 80B A3B Thinking	-	78.70%	MMLU-ProX	Yes	Source
Qwen	Qwen 3 VL 235B A22B Instruct	-	77.80%	-	Yes	-
Qwen	Qwen 3 VL 32B Thinking	-	77.20%	-	Yes	-
Qwen	Qwen 3 Next 80B A3B Instruct	-	76.70%	MMLU-ProX	Yes	Source
Qwen	Qwen 3.5 9B	02 Mar 2026	76.30%	-	Yes	Source
Qwen	Qwen 3 VL 30B A3B Thinking	-	76.10%	-	Yes	-
Qwen	Qwen 3 VL 32B Instruct	-	73.40%	-	Yes	-
Qwen	Qwen 3.5 4B	02 Mar 2026	71.50%	-	Yes	Source
Qwen	Qwen 3 VL 30B A3B Instruct	-	70.90%	-	Yes	-
Qwen	Qwen 3 8B	-	70.70%	inferred high-confidence family alias from qwen3-vl-8b-thinking (score=0.4600; benches=50)	Yes	-
Qwen	Qwen 3 Embedding 8B	-	70.70%	inferred family alias from qwen3-vl-8b-thinking (score=0.3850; benches=50)	Yes	-
Qwen	Qwen 3 Guard Gen 8B	-	70.70%	inferred family alias from qwen3-vl-8b-thinking (score=0.3400; benches=50)	Yes	-
Qwen	Qwen 3 Guard Stream 8B	-	70.70%	inferred family alias from qwen3-vl-8b-thinking (score=0.3371; benches=50)	Yes	-
Qwen	Qwen 3 Reranker 8B	-	70.70%	inferred family alias from qwen3-vl-8b-thinking (score=0.3850; benches=50)	Yes	-
Qwen	Qwen 3 VL 8B Thinking	-	70.70%	-	Yes	-
Qwen	Qwen 3 VL Embedding 8B	-	70.70%	inferred high-confidence family alias from qwen3-vl-8b-thinking (score=0.5232; benches=50)	Yes	-
Qwen	Qwen 3 VL Reranker 8B	-	70.70%	inferred high-confidence family alias from qwen3-vl-8b-thinking (score=0.5275; benches=50)	Yes	-
Qwen	Qwen 3 VL 8B Instruct	-	65.40%	-	Yes	-
Qwen	Qwen 3 4B Instruct 2507	-	65%	inferred family alias from qwen3-vl-4b-thinking (score=0.3462; benches=48)	Yes	-
Qwen	Qwen 3 4B SafeRL	-	65%	inferred family alias from qwen3-vl-4b-thinking (score=0.3850; benches=48)	Yes	-
Qwen	Qwen 3 4B Thinking 2507	-	65%	inferred family alias from qwen3-vl-4b-thinking (score=0.3462; benches=48)	Yes	-
Qwen	Qwen 3 Embedding 4B	-	65%	inferred family alias from qwen3-vl-4b-thinking (score=0.3850; benches=48)	Yes	-
Qwen	Qwen 3 Guard Gen 4B	-	65%	inferred family alias from qwen3-vl-4b-thinking (score=0.3400; benches=48)	Yes	-
Qwen	Qwen 3 Guard Stream 4B	-	65%	inferred family alias from qwen3-vl-4b-thinking (score=0.3371; benches=48)	Yes	-
Qwen	Qwen 3 Reranker 4B	-	65%	inferred family alias from qwen3-vl-4b-thinking (score=0.3850; benches=48)	Yes	-
Qwen	Qwen 3 VL 4B Thinking	-	65%	-	Yes	-
Nvidia	Nemotron Nano 3 30B A3B	15 Dec 2025	59.50%	-	Yes	Source
Qwen	Qwen 3 VL 4B Instruct	-	59.40%	-	Yes	-
Qwen	Qwen 3.5 2B	02 Mar 2026	52.30%	-	Yes	Source
Qwen	Qwen 3.5 0.8B	02 Mar 2026	34.60%	-	Yes	Source

Average Score

Score Range

Leading Model

MMLU-ProX

Recorded Results

Average Score

Score Range

Leading Model

Models Using This Benchmark