Science qa — Epoch AI

Science qa — Epoch AI is run by Epoch AI. It has ranked 26 models, of which the catalogue holds 12, scoring from 36.19 to 91.3 on Score.

measured by Epoch AI · the board itself

Place	Model	Metric	Score
1stof 26	Phi-3.5-vision-instruct	Score	91.3
2ndof 26	GPT-4o (2024-05-13)	Score	88.5
5thof 26	blip2-opt-2.7b	Score	74.17
10thof 26	llava-v1.6-mistral-7b	Score	72.8
12thof 26	Claude 3 Haiku	Score	72
13thof 26	llava-v1.6-vicuna-7b	Score	70.6
17thof 26	Qwen-VL-Chat	Score	68.2
19thof 26	llava-v1.5-7b	Score	66.8
23rdof 26	Llama-2-13b	Score	55.78
24thof 26	llama-13b	Score	43.33
25thof 26	Llama-2-7b	Score	43.08
26thof 26	llama-7b	Score	36.19

Place

Model

Metric

Score

1stof 26

Score

91.3

2ndof 26

Score

88.5