MMStar Benchmark

MMStar Benchmark | Phaseo

Models Using This Benchmark

Organisation	Model	Reported	Top Score	Info	Self Reported	Source
Qwen	Qwen 3.6 Plus	01 Apr 2026	83.30%	-	Yes	Source
ByteDance	Seed 2.0 Pro	14 Feb 2026	83%	Seed2 official benchmark table \| MMStar	Yes	Source
Qwen	Qwen 3.5 122B A10B	24 Feb 2026	82.90%	-	Yes	Source
Qwen	Qwen 3.5 35B A3B	24 Feb 2026	81.90%	-	Yes	Source
Qwen	Qwen 3.5 27B	24 Feb 2026	81%	-	Yes	Source
Qwen	Qwen 3.5 Flash	23 Feb 2026	81%	inferred family alias from qwen3.5-27b (score=0.4147; benches=81)	Yes	Source
ByteDance	Seed 2.0 Lite	14 Feb 2026	80.70%	Seed2 official benchmark table \| MMStar	Yes	Source
Qwen	Qwen 3 VL 32B Thinking	-	79.40%	-	Yes	-
ByteDance	Seed 2.0 Mini	14 Feb 2026	79.10%	Seed2 official benchmark table \| MMStar	Yes	Source
Qwen	Qwen 3 VL 235B A22B Thinking	-	78.70%	-	Yes	-
Qwen	Qwen 3 VL 235B A22B Instruct	-	78.40%	-	Yes	-
Qwen	Qwen 3 VL 32B Instruct	-	77.70%	-	Yes	-
Qwen	Qwen 3 VL 30B A3B Thinking	-	75.50%	-	Yes	-
Qwen	Qwen 3 8B	-	75.30%	inferred high-confidence family alias from qwen3-vl-8b-thinking (score=0.4600; benches=50)	Yes	-
Qwen	Qwen 3 Embedding 8B	-	75.30%	inferred family alias from qwen3-vl-8b-thinking (score=0.3850; benches=50)	Yes	-
Qwen	Qwen 3 Guard Gen 8B	-	75.30%	inferred family alias from qwen3-vl-8b-thinking (score=0.3400; benches=50)	Yes	-
Qwen	Qwen 3 Guard Stream 8B	-	75.30%	inferred family alias from qwen3-vl-8b-thinking (score=0.3371; benches=50)	Yes	-
Qwen	Qwen 3 Reranker 8B	-	75.30%	inferred family alias from qwen3-vl-8b-thinking (score=0.3850; benches=50)	Yes	-
Qwen	Qwen 3 VL 8B Thinking	-	75.30%	-	Yes	-
Qwen	Qwen 3 VL Embedding 8B	-	75.30%	inferred high-confidence family alias from qwen3-vl-8b-thinking (score=0.5232; benches=50)	Yes	-
Qwen	Qwen 3 VL Reranker 8B	-	75.30%	inferred high-confidence family alias from qwen3-vl-8b-thinking (score=0.5275; benches=50)	Yes	-
Qwen	Qwen 3 4B Instruct 2507	-	73.20%	inferred family alias from qwen3-vl-4b-thinking (score=0.3462; benches=48)	Yes	-
Qwen	Qwen 3 4B SafeRL	-	73.20%	inferred family alias from qwen3-vl-4b-thinking (score=0.3850; benches=48)	Yes	-
Qwen	Qwen 3 4B Thinking 2507	-	73.20%	inferred family alias from qwen3-vl-4b-thinking (score=0.3462; benches=48)	Yes	-
Qwen	Qwen 3 Embedding 4B	-	73.20%	inferred family alias from qwen3-vl-4b-thinking (score=0.3850; benches=48)	Yes	-
Qwen	Qwen 3 Guard Gen 4B	-	73.20%	inferred family alias from qwen3-vl-4b-thinking (score=0.3400; benches=48)	Yes	-
Qwen	Qwen 3 Guard Stream 4B	-	73.20%	inferred family alias from qwen3-vl-4b-thinking (score=0.3371; benches=48)	Yes	-
Qwen	Qwen 3 Reranker 4B	-	73.20%	inferred family alias from qwen3-vl-4b-thinking (score=0.3850; benches=48)	Yes	-
Qwen	Qwen 3 VL 4B Thinking	-	73.20%	-	Yes	-
Qwen	Qwen 3 VL 30B A3B Instruct	-	72.10%	-	Yes	-
Qwen	Qwen 3 VL 8B Instruct	-	70.90%	-	Yes	-
Qwen	Qwen 2.5 Math 72B	-	70.80%	inferred high-confidence family alias from qwen2.5-vl-72b (score=0.4700; benches=30)	Yes	Source
Qwen	Qwen 2.5 Math PRM 72B	-	70.80%	inferred family alias from qwen2.5-vl-72b (score=0.4050; benches=30)	Yes	Source
Qwen	Qwen 2.5 Math RM 72B	-	70.80%	inferred family alias from qwen2.5-vl-72b (score=0.4092; benches=30)	Yes	Source
Qwen	Qwen 2.5 VL 72B Instruct	26 Jan 2025	70.80%	-	Yes	Source
Qwen	Qwen 3 VL 4B Instruct	-	69.80%	-	Yes	-
Qwen	Qwen 2.5 Coder 32B Instruct	-	69.50%	inferred high-confidence family alias from qwen2.5-vl-32b (score=0.4641; benches=28)	Yes	Source
Qwen	Qwen 2.5 VL 32B Instruct	24 Mar 2025	69.50%	-	Yes	Source
Qwen	Qwen 2.5 VL 3B Instruct	-	69.50%	inferred high-confidence family alias from qwen2.5-vl-32b (score=0.4914; benches=28)	Yes	Source
Qwen	Qwen 2.5 Coder 3B	-	64%	inferred family alias from qwen2.5-omni-7b (score=0.3000; benches=45)	Yes	Source
Qwen	Qwen 2.5 Coder 7B	-	64%	inferred high-confidence family alias from qwen2.5-omni-7b (score=0.4700; benches=45)	Yes	Source
Qwen	Qwen 2.5 Math 7B	-	64%	inferred high-confidence family alias from qwen2.5-omni-7b (score=0.4767; benches=45)	Yes	Source
Qwen	Qwen 2.5 Math 7B PRM800K	-	64%	inferred family alias from qwen2.5-omni-7b (score=0.3696; benches=45)	Yes	Source
Qwen	Qwen 2.5 Math PRM 7B	-	64%	inferred family alias from qwen2.5-omni-7b (score=0.4092; benches=45)	Yes	Source
Qwen	Qwen 2.5 Omni 3B	-	64%	inferred high-confidence family alias from qwen2.5-omni-7b (score=0.4933; benches=45)	Yes	Source
Qwen	Qwen 2.5 Omni 7B	-	64%	-	Yes	Source
Qwen	Qwen 2.5 0.5B	-	63.90%	inferred family alias from qwen2.5-vl-7b (score=0.3131; benches=32)	Yes	Source
Qwen	Qwen 2.5 1.5B	-	63.90%	inferred family alias from qwen2.5-vl-7b (score=0.3131; benches=32)	Yes	Source
Qwen	Qwen 2.5 3B	-	63.90%	inferred family alias from qwen2.5-vl-7b (score=0.3531; benches=32)	Yes	Source
Qwen	Qwen 2.5 VL 7B Instruct	-	63.90%	-	Yes	Source
DeepSeek	DeepSeek VL2	13 Dec 2024	61.30%	-	Yes	Source
DeepSeek	DeepSeek VL2 Small	13 Dec 2024	57%	-	Yes	Source
DeepSeek	DeepSeek VL2 Tiny	13 Dec 2024	45.90%	-	Yes	Source

Average Score

Score Range

Leading Model

MMStar

Recorded Results

Average Score

Score Range

Leading Model

Models Using This Benchmark