RealWorldQA

AI benchmark results and model performance

Type: percentageSpatial Reasoning

Recorded Results

Average Score

75.26%

Score Range

64.20% - 86%

Leading Model (lowest score)

64.20% - DeepSeek VL2 Tiny

Scores Over Time

Individual benchmark scores plotted by date.

Models Using This Benchmark

Organisation	Model	Reported	Top Score	Info	Self Reported	Source
DeepSeek	DeepSeek VL2 Tiny	13 Dec 2024	64.20%	-	Yes	Source
DeepSeek	DeepSeek VL2 Small	13 Dec 2024	65.40%	-	Yes	Source
DeepSeek	DeepSeek VL2	13 Dec 2024	68.40%	-	Yes	Source
SpaceXAI	Grok 1.5V	12 Apr 2024	68.70%	-	Yes	Source
Qwen	Qwen 2.5 Coder 3B	-	70.30%	inferred family alias from qwen2.5-omni-7b (score=0.3000; benches=45)	Yes	Source
Qwen	Qwen 2.5 Coder 7B	-	70.30%	inferred high-confidence family alias from qwen2.5-omni-7b (score=0.4700; benches=45)	Yes	Source
Qwen	Qwen 2.5 Math 7B	-	70.30%	inferred high-confidence family alias from qwen2.5-omni-7b (score=0.4767; benches=45)	Yes	Source
Qwen	Qwen 2.5 Math 7B PRM800K	-	70.30%	inferred family alias from qwen2.5-omni-7b (score=0.3696; benches=45)	Yes	Source
Qwen	Qwen 2.5 Math PRM 7B	-	70.30%	inferred family alias from qwen2.5-omni-7b (score=0.4092; benches=45)	Yes	Source
Qwen	Qwen 2.5 Omni 3B	-	70.30%	inferred high-confidence family alias from qwen2.5-omni-7b (score=0.4933; benches=45)	Yes	Source
Qwen	Qwen 2.5 Omni 7B	-	70.30%	-	Yes	Source
Qwen	Qwen 3 VL 4B Instruct	-	70.90%	-	Yes	-
Qwen	Qwen 3 VL 8B Instruct	-	71.50%	-	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 8B	-	73.50%	inferred high-confidence family alias from qwen3-vl-8b-thinking (score=0.4600; benches=50)	Yes	-
Qwen	Qwen 3 Guard Stream 8B	-	73.50%	inferred family alias from qwen3-vl-8b-thinking (score=0.3371; benches=50)	Yes	-
Qwen	Qwen 3 VL 8B Thinking	-	73.50%	-	Yes	-
Qwen	Qwen 3 VL Embedding 8B	-	73.50%	inferred high-confidence family alias from qwen3-vl-8b-thinking (score=0.5232; benches=50)	Yes	-
Qwen	Qwen 3 VL Reranker 8B	-	73.50%	inferred high-confidence family alias from qwen3-vl-8b-thinking (score=0.5275; benches=50)	Yes	-
Qwen	Qwen 3 VL 30B A3B Instruct	-	73.70%	-	Yes	-
Qwen	Qwen 3 VL 30B A3B Thinking	-	77.40%	-	Yes	-
Qwen	Qwen 2 VL 2B	-	77.80%	inferred family alias from qwen2-vl-72b (score=0.4083; benches=15)	Yes	Source
Qwen	Qwen 2 VL 72B	-	77.80%	-	Yes	Source
Qwen	Qwen 3 VL 32B Thinking	-	78.40%	-	Yes	-
Qwen	Qwen 3 VL 32B Instruct	-	79%	-	Yes	-
Qwen	Qwen 3 VL 235B A22B Instruct	-	79.30%	-	Yes	-
Qwen	Qwen 3 VL 235B A22B Thinking	-	81.30%	-	Yes	-
ByteDance	Seed 2.0 Mini	14 Feb 2026	81.60%	Seed2 official benchmark table \| RealWorldQA	Yes	Source
ByteDance	Seed 2.0 Lite	14 Feb 2026	81.70%	Seed2 official benchmark table \| RealWorldQA	Yes	Source
Qwen	Qwen 3.5 27B	24 Feb 2026	83.70%	-	Yes	Source
Qwen	Qwen 3.5 Flash	23 Feb 2026	83.70%	inferred family alias from qwen3.5-27b (score=0.4147; benches=81)	Yes	Source
Qwen	Qwen 3.5 35B A3B	24 Feb 2026	84.10%	-	Yes	Source
Qwen	Qwen 3.5 122B A10B	24 Feb 2026	85.10%	-	Yes	Source
Qwen	Qwen 3.6 Plus	01 Apr 2026	85.40%	-	Yes	Source
Qwen	Qwen 3.8 27B	14 Aug 2026	85.90%	-	Yes	Source
ByteDance	Seed 2.0 Pro	14 Feb 2026	86%	Seed2 official benchmark table \| RealWorldQA	Yes	Source