SciTaste-Bench
Models judge the research value of papers in six disciplines. The benchmark compares their judgments with a reference index built from seven dimensions.
- Released
- October 2026
- Kind
- Research science
- Level
- Expert
- Scored
- Checked automatically
- Answers
- Prediction
- Tasks
- All public
- Measures
- Capability
- Language
- English
- Interaction
- Agentic
- Environment
- APIs and tools, Browser
- Items from
- Real work and use
- Refresh
- One fixed set
- Run by
- Anyone
- In views
- Agents, Biology and medicine, Chemistry and materials
- Strongest model tested
- Claude Fable 5 #12 on the capability index
- Last model added
- August 2026
- Scores refreshed
- November 20th, 2026
Qwen 3.8 Max leads with 28.0 among models released by August 2026
Best score per model from the benchmark's own leaderboard, read on November 20th, 2026. Where a model ran with several settings or harnesses, we keep its best. Every run is in the next chart.
The newest model with a score here came out in August 2026, 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 14 more
- 1
Qwen 3.8 Max ADS tools 28.0
- 2
Qwen3.7-Plus ADS tools 27.8
- 3
Qwen3.7-Max ADS tools 27.6
- 4 Claude Fable 5 ADS tools 26.8
- 5 Claude Opus 4.8 Native search 25.2
- 6 GPT-5.6 Sol Native search 22.2
- 7
Gemini 3.5 Flash ADS tools 19.9
Source: Official SciTaste-Bench leaderboard
| # | Model | Score | Setting | Date | Source |
|---|---|---|---|---|---|
| 1 | | 28.0 | ADS tools, DimAgent-v3 | Aug 2026 | Official leaderboard |
| 2 | | 27.8 | ADS tools, DimAgent-v3 | Jun 2026 | Official leaderboard |
| 3 | | 27.6 | ADS tools, DimAgent-v3 | May 2026 | Official leaderboard |
| 4 | Claude Fable 5 | 26.8 | ADS tools, DimAgent-v3 | Jun 2026 | Official leaderboard |
| 5 | Claude Opus 4.8 | 25.2 | Native search, DimAgent-v3 | May 2026 | Official leaderboard |
| 6 | GPT-5.6 Sol | 22.2 | Native search, DimAgent-v3 | Jul 2026 | Official leaderboard |
| 7 | | 19.9 | ADS tools, DimAgent-v3 | May 2026 | Official leaderboard |
Rankings by part
We read 9 parts from the benchmark’s own leaderboard, each ranking models on one part of the test, such as Chemistry, Medicine and Mathematics. Qwen3.7-Plus 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.
Chemistry
7 models we track have a score here.
- 1
Qwen 3.8 Max 27.8
- 2
Qwen3.7-Plus 26.8
- 3
Qwen3.7-Max 26.4
- 4 Claude Opus 4.8 ↑1 25.5
- 5 Claude Fable 5 ↓1 25.5
- 6 GPT-5.6 Sol 24.2
- 7
Gemini 3.5 Flash 19.7
Medicine
7 models we track have a score here.
- 1
Qwen3.7-Plus ↑1 27.5
- 2
Qwen3.7-Max ↑1 26.5
- 3 Claude Fable 5 ↑1 25.9
- 4 Claude Opus 4.8 ↑1 25.5
- 5
Qwen 3.8 Max ↓4 25.3
- 6 GPT-5.6 Sol 24.6
- 7
Gemini 3.5 Flash 18.7
Mathematics
7 models we track have a score here.
- 1 Claude Opus 4.8 ↑4 28.4
- 2 Claude Fable 5 ↑2 28.3
- 3
Qwen3.7-Plus ↓1 28.1
- 4
Qwen 3.8 Max ↓3 27.7
- 5
Qwen3.7-Max ↓2 26.7
- 6 GPT-5.6 Sol 24.5
- 7
Gemini 3.5 Flash 18.9
Physics
7 models we track have a score here.
- 1
Qwen3.7-Max ↑2 35.2
- 2
Qwen3.7-Plus 34.7
- 3
Qwen 3.8 Max ↓2 34.7
- 4 Claude Fable 5 34.2
- 5 Claude Opus 4.8 32.1
- 6 GPT-5.6 Sol 29.1
- 7
Gemini 3.5 Flash 26.1
Biology
7 models we track have a score here.
- 1
Qwen3.7-Plus ↑1 26.3
- 2
Qwen3.7-Max ↑1 26.1
- 3
Qwen 3.8 Max ↓2 25.0
- 4 Claude Fable 5 24.8
- 5 Claude Opus 4.8 23.2
- 6 GPT-5.6 Sol 21.3
- 7
Gemini 3.5 Flash 19.8
Finance
7 models we track have a score here.
- 1
Qwen3.7-Plus ↑1 31.5
- 2
Qwen 3.8 Max ↓1 31.0
- 3 Claude Opus 4.8 ↑2 30.7
- 4
Qwen3.7-Max ↓1 30.0
- 5 Claude Fable 5 ↓1 30.0
- 6 GPT-5.6 Sol 26.7
- 7
Gemini 3.5 Flash 22.6
Method 2 score
7 models we track have a score here.
- 1
Qwen 3.8 Max 51.8
- 2
Qwen3.7-Plus 48.1
- 3
Qwen3.7-Max 47.7
- 4 Claude Fable 5 47.0
- 5 Claude Opus 4.8 40.7
- 6 GPT-5.6 Sol 28.7
- 7
Gemini 3.5 Flash 17.7
Spearman correlation
7 models we track have a score here.
- 1
Qwen3.7-Plus ↑1 0.57
- 2 Claude Fable 5 ↑2 0.56
- 3
Qwen3.7-Max 0.56
- 4
Qwen 3.8 Max ↓3 0.55
- 5 GPT-5.6 Sol ↑1 0.49
- 6
Gemini 3.5 Flash ↑1 0.49
- 7 Claude Opus 4.8 ↓2 0.46
Pearson correlation
7 models we track have a score here.
- 1
Qwen3.7-Plus ↑1 0.57
- 2
Qwen3.7-Max ↑1 0.56
- 3
Qwen 3.8 Max ↓2 0.56
- 4 Claude Fable 5 0.55
- 5
Gemini 3.5 Flash ↑2 0.51
- 6 GPT-5.6 Sol 0.49
- 7 Claude Opus 4.8 ↓2 0.45
Scores at each effort level
Models the leaderboard ran at two or more effort levels. We don’t have enough cost data here to say what the extra effort costs.
Source: Official SciTaste-Bench leaderboard
| Model | Effort | Harness | Score | Cost per task | Output tokens |
|---|---|---|---|---|---|
| ADS tools | DimAgent-v3 | 28.0 | |||
| ADS tools | DimAgent-v3 | 27.8 | |||
| Native search | DimAgent-v3 | 27.7 | |||
| ADS tools | DimAgent-v3 | 27.6 | |||
| Native search | DimAgent-v3 | 27.4 | |||
| Native search | DimAgent-v3 | 27.1 | |||
| Claude Fable 5 | ADS tools | DimAgent-v3 | 26.8 | ||
| Claude Opus 4.8 | Native search | DimAgent-v3 | 25.2 | ||
| Claude Fable 5 | Native search | DimAgent-v3 | 25.2 | ||
| Claude Fable 5 | Native Search | 24.6 | |||
| Claude Opus 4.8 | ADS tools | DimAgent-v3 | 23.7 | ||
| Native Search | 23.2 | ||||
| Claude Fable 5 | Pure QA | 22.3 | |||
| GPT-5.6 Sol | Native search | DimAgent-v3 | 22.2 | ||
| Claude Opus 4.8 | Native Search | 21.9 | |||
| GPT-5.6 Sol | Native Search | 21.1 | |||
| Pure QA | 21.1 | ||||
| Native Search | 20.2 | ||||
| Claude Opus 4.8 | Pure QA | 20.2 | |||
| ADS tools | DimAgent-v3 | 19.9 | |||
| Native Search | 19.8 | ||||
| Native search | DimAgent-v3 | 19.7 | |||
| GPT-5.6 Sol | Pure QA | 19.5 | |||
| Pure QA | 19.4 | ||||
| Pure QA | 19.3 | ||||
| Pure QA | 16.9 | ||||
| Native Search | 16.4 | ||||
| GPT-5.6 Sol | ADS tools | DimAgent-v3 | 9.50 |
Number one over time
Lab claims count from the day the card came out, independent runs from the model’s release.
Source: Official SciTaste-Bench leaderboard
Scores over time
All the scores we have, one dot each, so a model tested at several settings appears more than once.
Source: Official SciTaste-Bench 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
All 9 checks pass. Every benchmark
Catalog record 2 pass
- Pass Task access All tasks are public.
- Pass Confidence Checked against a primary source.
Scores 3 pass 2 notes
- Pass Independent runs Scores from the official leaderboard (7 entries).
- Pass Models scored 7 models we track.
- Pass Newest score The newest is from August 2026, for Qwen 3.8 Max.
- Note Frontier models None of the top 10 has a score. The strongest tested is Claude Fable 5 (#12).
- Note Headroom Scored as a plain number, with no fixed ceiling.
Leaderboard import 3 pass
- Pass Importer We read the board on huggingface.co: 28 rows for models we track 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.