math benchmark
SciCode
SciCode is a research coding benchmark curated by scientists that challenges language models to code solutions for scientific problems. It contains 338 subproblems decomposed from 80 challenging main problems across 16 natural science sub-fields including mathematics, physics, chemistry, biology, and materials science. Problems require knowledge recall, reasoning, and code synthesis skills.
Updated Aug 11, 2026
Models19
Model coverage19
MetricScore
EvidenceB
SciCode Ranking
Higher score ranks better on this benchmark.
| 1 | BY | 59.8% | 100.0% | 19 | C | |
| 2 | GO | 59.0% | 94.4% | 19 | C | |
| 3 | BY | 57.8% | 88.9% | 19 | C | |
| 4 | AC | 53.5% | 83.3% | 19 | C | |
| 5 | MA | 52.2% | 77.8% | 19 | C | |
| 6 | AC | 51.3% | 72.2% | 19 | C | |
| 7 | MA | 48.7% | 66.7% | 19 | C | |
| 8 | MA | 44.8% | 61.1% | 19 | C | |
| 9 | NV | 44.6% | 55.6% | 19 | C | |
| 10 | ME | 43.6% | 50.0% | 19 | C | |
| 11 | NV | 42.0% | 44.4% | 19 | C | |
| 12 | ZA | 41.7% | 38.9% | 19 | C | |
| 13 | MI | 39.0% | 33.3% | 19 | C | |
| 14 | CO | 38.2% | 27.8% | 19 | C | |
| 15 | CO | 38.0% | 22.2% | 19 | C | |
| 16 | IN | 38.0% | 16.7% | 19 | C | |
| 17 | ZA | 37.3% | 11.1% | 19 | C | |
| 18 | MI | 36.0% | 5.6% | 19 | C | |
| 19 | NV | 33.3% | 0.0% | 19 | C |
SciCode Score Distribution
A closer view of the leading scores on this benchmark.
SciCode Highlights
The leading models and scores on this benchmark.
What is SciCode?
What SciCode measures and how its scores work.
SciCode is a research coding benchmark curated by scientists that challenges language models to code solutions for scientific problems. It contains 338 subproblems decomposed from 80 challenging main problems across 16 natural science sub-fields including mathematics, physics, chemistry, biology, and materials science. Problems require knowledge recall, reasoning, and code synthesis skills.
Scores are shown in ratio. This benchmark is not independently verified and has an evidence level of B.
- Family
- SciCode
- Modality
- text
- Primary category
- math
- Score direction
- higher
- LLMBoard eligible
- Yes
- Evaluation key
- scicode|llm-stats-current
Benchmark scores retain their original unit. Overall score eligibility is shown separately.
FAQ
Common questions about SciCode.
Which model scores highest on SciCode?
Seed 2.1 Pro is currently ranked first with 59.8%.
What does SciCode measure?
SciCode is a research coding benchmark curated by scientists that challenges language models to code solutions for scientific problems. It contains 338 subproblems decomposed from 80 challenging main problems across 16 natural science sub-fields including mathematics, physics, chemistry, biology, and materials science. Problems require knowledge recall, reasoning, and code synthesis skills.
Is a higher score better?
Yes. Higher values rank better for this benchmark.
How many models are compared?
19 model results are currently shown.
Does this benchmark affect the overall score?
Yes. This benchmark can contribute to the current LLMBoard capability score.