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 7, 2026
Published models18
Registry coverage18
MetricScore
EvidenceB
SciCode leaderboard
Sorted by the source-provided rank. Higher score is better according to the registry.
| 1 | BY | 59.8% | 100.0% | 18 | C | |
| 2 | GO | 59.0% | 94.1% | 18 | C | |
| 3 | BY | 57.8% | 88.2% | 18 | C | |
| 4 | AC | 53.5% | 82.3% | 18 | C | |
| 5 | MA | 52.2% | 76.5% | 18 | C | |
| 6 | AC | 51.3% | 70.6% | 18 | C | |
| 7 | MA | 48.7% | 64.7% | 18 | C | |
| 8 | MA | 44.8% | 58.8% | 18 | C | |
| 9 | NV | 44.6% | 52.9% | 18 | C | |
| 10 | NV | 42.0% | 47.1% | 18 | C | |
| 11 | ZA | 41.7% | 41.2% | 18 | C | |
| 12 | MI | 39.0% | 35.3% | 18 | C | |
| 13 | CO | 38.2% | 29.4% | 18 | C | |
| 14 | CO | 38.0% | 23.5% | 18 | C | |
| 15 | IN | 38.0% | 17.6% | 18 | C | |
| 16 | ZA | 37.3% | 11.8% | 18 | C | |
| 17 | MI | 36.0% | 5.9% | 18 | C | |
| 18 | NV | 33.3% | 0.0% | 18 | C |
Score distribution
Top published rows on the benchmark's original scale.
SciCode highlights
The top published results on this benchmark's own scale.
What is SciCode?
Definition and scoring fields from the benchmark registry.
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. The current registry marks this benchmark as not independently verified with evidence level B.
- Family
- SciCode
- Modality
- text
- Primary category
- math
- Score direction
- higher
- LLMBoard eligible
- Yes
- Evaluation key
- scicode|llm-stats-current
Source-native results are preserved. Eligibility for the overall LLMBoard score is a separate policy decision.
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?
18 unique published model results are currently shown.
Does this benchmark affect the overall score?
This benchmark is marked as eligible for the current LLMBoard capability methodology.