Benchmarks
ScienceQA
Models scored
26
Best score
0.91
Metric
Score
ScienceQA is an AI evaluation tracked by Epoch AI, with 26 scored model versions on record. The scores here are reported by third parties rather than produced by Epoch AI running the evaluation itself.
Scored models span January 2022 to August 2024. The best score recorded is 0.91, measured as Score. Models come from Salesforce Research, OpenAI, Meta AI, Anthropic, Google DeepMind.
Full record
- Score metric
- Score
- Models scored
- 26
- Best score recorded
- 0.91
- Earliest model scored
- Jan 27, 2022
- Most recent model scored
- Aug 16, 2024
- Third-party reported
- Yes
Leaderboard
| 01 | Phi-3.5-vision-instruct | 0.91 | |
| 02 | gpt-4o-2024-05-13 | OpenAI | 0.89 |
| 03 | gemini-1.0-pro-vision | Google DeepMind | 0.8 |
| 04 | falcon-11B-vlm | 0.75 | |
| 05 | blip2-opt-2.7b | Salesforce Research | 0.74 |
| 06 | text-davinci-001 | OpenAI | 0.74 |
| 07 | llama3-llava-next-8b | 0.74 | |
| 08 | llava-v1.6-vicuna-13b | 0.74 | |
| 09 | llava-v1.6-mistral-7b | 0.73 | |
| 10 | MM1-7B-Chat | 0.73 | |
| 11 | claude-3-haiku-20240307 | Anthropic | 0.72 |
| 12 | llava-v1.6-vicuna-7b | 0.71 | |
| 13 | InternVL-Chat-ViT-6B-Vicuna-13B | 0.7 | |
| 14 | MM1-3B-Chat | 0.69 | |
| 15 | Qwen-VL-Chat | 0.68 | |
| 16 | llava-v1.5-7b | 0.67 | |
| 17 | InternVL-Chat-ViT-6B-Vicuna-7B | 0.66 | |
| 18 | instructblip-vicuna-13b | 0.63 | |
| 19 | instructblip-vicuna-7b | 0.6 | |
| 20 | Llama-2-13b | Meta AI | 0.56 |
| 21 | LLaMA-13B | Meta AI | 0.43 |
| 22 | Llama-2-7b | Meta AI | 0.43 |
| 23 | LLaMA-7B | Meta AI | 0.36 |
More benchmarks
SourceEpoch AI, 'AI Benchmarking Hub'. Published online at epoch.ai. Retrieved 2026-07-29 from https://epoch.ai/benchmarks. Licensed under CC BY 4.0.
← All benchmarks