SciKnowEval
Models answer scientific questions and apply knowledge in biology, chemistry, materials and physics. The suite uses automatic checks and model judges for different tasks.
- Made by
- Kehua Feng and colleagues
- Released
- June 2024
- Kind
- Science questions
- Level
- University
- Scored
- Graded by a model
- Answers
- Open-ended
- Tasks
- All public
- Measures
- Capability
- Language
- English
- Interaction
- Single turn
- Refresh
- One fixed set
- Run by
- Anyone
- In views
- Biology and medicine, Chemistry and materials
- Strongest model tested
- o4-mini #95 on the capability index
- Last model added
- May 2025
- Scores refreshed
- November 20th, 2026
o4-mini leads with 0.71 among models released by May 2025 o4-mini leads with 0.71 among models released by May 2025
Best score per entry from the benchmark's paper, read on November 20th, 2026. A paper’s table doesn’t add newer models. Where a model ran with several settings or harnesses, we keep its best. Names in gray aren't in our model list: fine-tunes, agent systems, products and older models. They don't count toward leaders. This view shows only models we track, each at its best.
The newest model with a score here came out in May 2025, and nobody has tested a newer one. Among the 20 highest-ranked models on the capability index, these have no score yet:
- GPT-6.2 Astra
- Claude Fable 5.2
-
Gemini 4 Carbon
- GPT-6 Astra
- and 16 more
- 0.71 o4-mini
- 0.70 o3-mini
- 0.68 GPT-4.1
- 0.66 DeepSeek-V3 (snapshot unspecified)
- 0.66
QwQ-32B
- 0.65 GPT-4o (snapshot unspecified)
- 0.65 DeepSeek-R1 (snapshot unspecified)
- 0.64
Qwen2.5-72B
- 0.63 Claude Sonnet 4
- 0.63
Llama 4 Scout
- 0.62
Qwen3-8B
- 0.62 GPT-4o-mini (snapshot unspecified)
- 0.56
Qwen2-7B-Instruct
- 0.52 ChemDFM-13B
- 0.48 ChemLLM-20B-Chat
- 0.48 Llama3-8B-Instruct
- 0.47 MolInst-Llama3-8B
- 0.40 SciGLM-6B
- 0.23 LlaSMol-Mistral-7B
- 1 o4-mini Zero-shot 0.71
- 2 o3-mini Zero-shot 0.70
- 3 GPT-4.1 Zero-shot 0.68
- 4 DeepSeek-V3 (snapshot unspecified) Zero-shot 0.66
- 5
QwQ-32B Zero-shot 0.66
- 6 GPT-4o (snapshot unspecified) Zero-shot 0.65
- 7 DeepSeek-R1 (snapshot unspecified) Zero-shot 0.65
- 8
Qwen2.5-72B Zero-shot 0.64
- 9 Claude Sonnet 4 Zero-shot, thinking 0.63
- 10
Llama 4 Scout Zero-shot 0.63
- 11
Qwen3-8B Zero-shot, thinking 0.62
- 12 GPT-4o-mini (snapshot unspecified) Zero-shot 0.62
- 13
Qwen2-7B-Instruct Zero-shot 0.56
- 14 ChemDFM-13B Zero-shot 0.52
- 15 ChemLLM-20B-Chat Zero-shot 0.48
- 16 Llama3-8B-Instruct Zero-shot 0.48
- 17 MolInst-Llama3-8B Zero-shot 0.47
- 18 SciGLM-6B Zero-shot 0.40
- 19 LlaSMol-Mistral-7B Zero-shot 0.23
Source: The SciKnowEval paper
- 1 o4-mini Zero-shot 0.71
- 2 o3-mini Zero-shot 0.70
- 3 GPT-4.1 Zero-shot 0.68
- 4
QwQ-32B Zero-shot 0.66
- 5
Qwen2.5-72B Zero-shot 0.64
- 6 Claude Sonnet 4 Zero-shot, thinking 0.63
- 7
Llama 4 Scout Zero-shot 0.63
- 8
Qwen3-8B Zero-shot, thinking 0.62
- 9
Qwen2-7B-Instruct Zero-shot 0.56
Source: The SciKnowEval paper
| # | Entry | Score | Setting | Date | Source |
|---|---|---|---|---|---|
| 1 | o4-mini | 0.71 | Zero-shot | Apr 2025 | Official leaderboard |
| 2 | o3-mini | 0.70 | Zero-shot | Jan 2025 | Official leaderboard |
| 3 | GPT-4.1 | 0.68 | Zero-shot | Apr 2025 | Official leaderboard |
| 4 | DeepSeek-V3 (snapshot unspecified) | 0.66 | Zero-shot | Official leaderboard | |
| 5 | | 0.66 | Zero-shot | Mar 2025 | Official leaderboard |
| 6 | GPT-4o (snapshot unspecified) | 0.65 | Zero-shot | Official leaderboard | |
| 7 | DeepSeek-R1 (snapshot unspecified) | 0.65 | Zero-shot | Official leaderboard | |
| 8 | | 0.64 | Zero-shot | Sep 2024 | Official leaderboard |
| 9 | Claude Sonnet 4 | 0.63 | Zero-shot, thinking | May 2025 | Official leaderboard |
| 10 | | 0.63 | Zero-shot | Apr 2025 | Official leaderboard |
| 11 | | 0.62 | Zero-shot, thinking | Apr 2025 | Official leaderboard |
| 12 | GPT-4o-mini (snapshot unspecified) | 0.62 | Zero-shot | Official leaderboard | |
| 13 | | 0.56 | Zero-shot | Jun 2024 | Official leaderboard |
| 14 | ChemDFM-13B | 0.52 | Zero-shot | Official leaderboard | |
| 15 | ChemLLM-20B-Chat | 0.48 | Zero-shot | Official leaderboard | |
| 16 | Llama3-8B-Instruct | 0.48 | Zero-shot | Official leaderboard | |
| 17 | MolInst-Llama3-8B | 0.47 | Zero-shot | Official leaderboard | |
| 18 | SciGLM-6B | 0.40 | Zero-shot | Official leaderboard | |
| 19 | LlaSMol-Mistral-7B | 0.23 | Zero-shot | Official leaderboard |
| # | Model | Score | Setting | Date | Source |
|---|---|---|---|---|---|
| 1 | o4-mini | 0.71 | Zero-shot | Apr 2025 | Official leaderboard |
| 2 | o3-mini | 0.70 | Zero-shot | Jan 2025 | Official leaderboard |
| 3 | GPT-4.1 | 0.68 | Zero-shot | Apr 2025 | Official leaderboard |
| 4 | | 0.66 | Zero-shot | Mar 2025 | Official leaderboard |
| 5 | | 0.64 | Zero-shot | Sep 2024 | Official leaderboard |
| 6 | Claude Sonnet 4 | 0.63 | Zero-shot, thinking | May 2025 | Official leaderboard |
| 7 | | 0.63 | Zero-shot | Apr 2025 | Official leaderboard |
| 8 | | 0.62 | Zero-shot, thinking | Apr 2025 | Official leaderboard |
| 9 | | 0.56 | Zero-shot | Jun 2024 | Official leaderboard |
Rankings by part
We read 9 parts from the benchmark’s paper in 2 groups: level and domain. Each ranks models on one part of the test. o4-mini leads 5 of them, and 3 other models lead the rest. Arrows show how far a model moved from its overall rank among the models we track.
L1
9 models we track have a score here.
- 1 GPT-4.1 ↑2 0.86
- 2 o3-mini 0.86
- 3 o4-mini ↓2 0.86
- 4
Qwen2.5-72B ↑1 0.83
- 5
QwQ-32B ↓1 0.82
- 6
Llama 4 Scout ↑1 0.82
- 7
Qwen3-8B ↑1 0.80
- 8 Claude Sonnet 4 ↓2 0.77
- 9
Qwen2-7B-Instruct 0.76
L2
9 models we track have a score here.
- 1 Claude Sonnet 4 ↑5 0.85
- 2 GPT-4.1 ↑1 0.84
- 3 o4-mini ↓2 0.84
- 4
QwQ-32B 0.84
- 5 o3-mini ↓3 0.84
- 6
Qwen2.5-72B ↓1 0.82
- 7
Qwen3-8B ↑1 0.82
- 8
Qwen2-7B-Instruct ↑1 0.79
- 9
Llama 4 Scout ↓2 0.79
L3
9 models we track have a score here.
- 1 o3-mini ↑1 0.60
- 2 o4-mini ↓1 0.59
- 3
QwQ-32B ↑1 0.57
- 4
Llama 4 Scout ↑3 0.54
- 5 GPT-4.1 ↓2 0.53
- 6
Qwen2.5-72B ↓1 0.48
- 7 Claude Sonnet 4 ↓1 0.46
- 8
Qwen3-8B 0.45
- 9
Qwen2-7B-Instruct 0.38
L4
9 models we track have a score here.
- 1 o4-mini 0.77
- 2 o3-mini 0.70
- 3 GPT-4.1 0.69
- 4 Claude Sonnet 4 ↑2 0.69
- 5
Qwen2.5-72B 0.68
- 6
Qwen3-8B ↑2 0.65
- 7
QwQ-32B ↓3 0.64
- 8
Llama 4 Scout ↓1 0.64
- 9
Qwen2-7B-Instruct 0.63
L5
9 models we track have a score here.
- 1 o4-mini 0.49
- 2 o3-mini 0.49
- 3 GPT-4.1 0.47
- 4 Claude Sonnet 4 ↑2 0.42
- 5
QwQ-32B ↓1 0.42
- 6
Qwen2.5-72B ↓1 0.38
- 7
Llama 4 Scout 0.38
- 8
Qwen3-8B 0.37
- 9
Qwen2-7B-Instruct 0.26
Biology
9 models we track have a score here.
- 1 o4-mini 0.63
- 2 GPT-4.1 ↑1 0.61
- 3 o3-mini ↓1 0.60
- 4
Qwen2.5-72B ↑1 0.57
- 5 Claude Sonnet 4 ↑1 0.57
- 6
Llama 4 Scout ↑1 0.56
- 7
Qwen3-8B ↑1 0.56
- 8
QwQ-32B ↓4 0.55
- 9
Qwen2-7B-Instruct 0.53
Chemistry
9 models we track have a score here.
- 1 o3-mini ↑1 0.69
- 2 o4-mini ↓1 0.68
- 3
QwQ-32B ↑1 0.64
- 4 GPT-4.1 ↓1 0.63
- 5 Claude Sonnet 4 ↑1 0.63
- 6
Qwen3-8B ↑2 0.60
- 7
Llama 4 Scout 0.60
- 8
Qwen2.5-72B ↓3 0.59
- 9
Qwen2-7B-Instruct 0.53
Material
9 models we track have a score here.
- 1 o4-mini 0.74
- 2 o3-mini 0.72
- 3
QwQ-32B ↑1 0.72
- 4 GPT-4.1 ↓1 0.70
- 5 Claude Sonnet 4 ↑1 0.69
- 6
Llama 4 Scout ↑1 0.67
- 7
Qwen2.5-72B ↓2 0.65
- 8
Qwen3-8B 0.60
- 9
Qwen2-7B-Instruct 0.55
Physics
9 models we track have a score here.
- 1 o4-mini 0.89
- 2 o3-mini 0.89
- 3
QwQ-32B ↑1 0.87
- 4 GPT-4.1 ↓1 0.85
- 5
Qwen2.5-72B 0.85
- 6 Claude Sonnet 4 0.84
- 7
Llama 4 Scout 0.83
- 8
Qwen3-8B 0.82
- 9
Qwen2-7B-Instruct 0.76
4 models have held number one in the benchmark’s paper since its release in June 2024
Lab claims count from the day the card came out, independent runs from the model’s release.
Source: The SciKnowEval paper
Scores over time
All the scores we have, one dot each, so a model tested at several settings appears more than once.
Source: Official SciKnowEval leaderboard
No lab model card reports it yet
We read the headline results of 540 launch posts from 57 labs, and none of them list this benchmark.
More in Science and reasoning
- GPQA Diamond In use 198 graduate-level 'Google-proof' multiple-choice questions in biology, chemistry and physics.
Gemini 4 Carbon 97%
- Terminal-Bench-Science In use 70 scientific research-workflow tasks across five science areas, run in Harbor containers. Hidden tests give the resolution rate. GPT-6 Astra 68.1%
- BioMysteryBench In use Bioinformatics problems on anonymized real datasets that require real analysis with tools. Rubric-graded, with human-solvable and human-difficult splits. Claude Fable 5.2 83.7%
- FrontierScience In use Models solve physics, chemistry, and biology problems in an Olympiad track graded by short answer and a Research track graded by 10-point rubrics.
Intern-S2-397B 87%
- HiPhO In use Models solve theory problems from 2024 and 2025 physics olympiad exams, some with diagrams. Official marking schemes grade answers and steps, mapped to medal levels.
Gemini 3 Pro 78.3%
- LABBench2 In use 11 biology research sub-tasks (literature, figures, tables, patents, trials and more) with tool access. Macro-averaged accuracy. GPT-5.2 60.9%
Data audit
8 of 9 checks pass, 1 warning. Look first at the Newest score check. The newest came 18 months ago, in May 2025, for Claude Sonnet 4. Every benchmark
Catalog record 2 pass
- Pass Task access All tasks are public.
- Pass Confidence Checked against a primary source.
Scores 1 warning 2 pass 2 notes
- Pass Independent runs Scores from the benchmark’s paper (19 entries).
- Pass Models scored 9 models we track, plus 10 other entries on its leaderboard.
- Warning Newest score The newest came 18 months ago, in May 2025, for Claude Sonnet 4.
- Note Frontier models None of the top 10 has a score. The strongest tested is o4-mini (#95).
- Note Headroom Scored as a plain number, with no fixed ceiling.
Leaderboard import 3 pass
- Pass Importer We read the board on arxiv.org: 10 rows for models we track and 10 other entries at the last read.
- Pass Last read November 20th, 2026.
- Pass Read worked The last read worked.
Consistency 1 pass 1 note
- Note Lab claims No model has both a lab number and an independent run.
- Pass Known issues No published audit found problems.