TextVQA

AI benchmark results and model performance

Type: percentageImage To Text

Recorded Results

Average Score

81.95%

Score Range

72% - 85.50%

Leading Model (lowest score)

72% - Phi 3 Vision 128K Instruct

Scores Over Time

Individual benchmark scores plotted by date.

Models Using This Benchmark

Organisation	Model	Reported	Top Score	Info	Self Reported	Source
Microsoft	Phi 3 Vision 128K Instruct	-	72%	inferred family alias from phi-3.5-vision-instruct (score=0.3495; benches=9)	Yes	Source
Microsoft	Phi 3.5 vision instruct	23 Aug 2024	72%	-	Yes	Source
Meta	Llama 3.2 90B Vision Instruct	-	73.50%	inferred high-confidence family alias from llama-3.2-90b-instruct (score=0.5873; benches=13)	Yes	Source
Microsoft	Phi 4 multimodal instruct	01 Feb 2025	75.60%	-	Yes	Source
SpaceXAI	Grok 1.5V	12 Apr 2024	78.10%	-	Yes	Source
DeepSeek	DeepSeek VL2 Tiny	13 Dec 2024	80.70%	-	Yes	Source
DeepSeek	DeepSeek VL2 Small	13 Dec 2024	83.40%	-	Yes	Source
DeepSeek	DeepSeek VL2	13 Dec 2024	84.20%	-	Yes	Source
Qwen	Qwen 2.5 Coder 3B	-	84.40%	inferred family alias from qwen2.5-omni-7b (score=0.3000; benches=45)	Yes	Source
Qwen	Qwen 2.5 Coder 7B	-	84.40%	inferred high-confidence family alias from qwen2.5-omni-7b (score=0.4700; benches=45)	Yes	Source
Qwen	Qwen 2.5 Math 7B	-	84.40%	inferred high-confidence family alias from qwen2.5-omni-7b (score=0.4767; benches=45)	Yes	Source
Qwen	Qwen 2.5 Math 7B PRM800K	-	84.40%	inferred family alias from qwen2.5-omni-7b (score=0.3696; benches=45)	Yes	Source
Qwen	Qwen 2.5 Math PRM 7B	-	84.40%	inferred family alias from qwen2.5-omni-7b (score=0.4092; benches=45)	Yes	Source
Qwen	Qwen 2.5 Omni 3B	-	84.40%	inferred high-confidence family alias from qwen2.5-omni-7b (score=0.4933; benches=45)	Yes	Source
Qwen	Qwen 2.5 Omni 7B	-	84.40%	-	Yes	Source
Qwen	Qwen 2.5 0.5B	-	84.90%	inferred family alias from qwen2.5-vl-7b (score=0.3131; benches=32)	Yes	Source
Qwen	Qwen 2.5 1.5B	-	84.90%	inferred family alias from qwen2.5-vl-7b (score=0.3131; benches=32)	Yes	Source
Qwen	Qwen 2.5 3B	-	84.90%	inferred family alias from qwen2.5-vl-7b (score=0.3531; benches=32)	Yes	Source
Qwen	Qwen 2.5 VL 7B Instruct	-	84.90%	-	Yes	Source
Qwen	Qwen 2 VL 2B	-	85.50%	inferred family alias from qwen2-vl-72b (score=0.4083; benches=15)	Yes	Source
Qwen	Qwen 2 VL 72B	-	85.50%	-	Yes	Source