multimodal benchmark
LifeSciBench
LifeSciBench is an expert-authored, expert-reviewed benchmark of 750 open-ended life-science research tasks spanning seven workflows and seven biological domains. Responses are graded against 19,020 physician- and scientist-written rubric criteria rather than multiple-choice answers, and most tasks require interpreting attached artifacts such as figures, PDFs, and sequence files.
Updated Aug 7, 2026
Published models3
Registry coverage3
MetricScore
EvidenceB
LifeSciBench leaderboard
Sorted by the source-provided rank. Higher score is better according to the registry.
| 1 | OP | 59.9% | 100.0% | 3 | C | |
| 2 | OP | 56.0% | 50.0% | 3 | C | |
| 3 | OP | 51.2% | 0.0% | 3 | C |
Score distribution
Top published rows on the benchmark's original scale.
LifeSciBench highlights
The top published results on this benchmark's own scale.
What is LifeSciBench?
Definition and scoring fields from the benchmark registry.
LifeSciBench is an expert-authored, expert-reviewed benchmark of 750 open-ended life-science research tasks spanning seven workflows and seven biological domains. Responses are graded against 19,020 physician- and scientist-written rubric criteria rather than multiple-choice answers, and most tasks require interpreting attached artifacts such as figures, PDFs, and sequence files.
Scores are shown in ratio. The current registry marks this benchmark as not independently verified with evidence level B.
- Family
- LifeSciBench
- Modality
- multimodal
- Primary category
- multimodal
- Score direction
- higher
- LLMBoard eligible
- No
- Evaluation key
- lifescibench|llm-stats-current
Source-native results are preserved. Eligibility for the overall LLMBoard score is a separate policy decision.
FAQ
Common questions about LifeSciBench.
Which model scores highest on LifeSciBench?
GPT-5.6 Sol is currently ranked first with 59.9%.
What does LifeSciBench measure?
LifeSciBench is an expert-authored, expert-reviewed benchmark of 750 open-ended life-science research tasks spanning seven workflows and seven biological domains. Responses are graded against 19,020 physician- and scientist-written rubric criteria rather than multiple-choice answers, and most tasks require interpreting attached artifacts such as figures, PDFs, and sequence files.
Is a higher score better?
Yes. Higher values rank better for this benchmark.
How many models are compared?
3 unique published model results are currently shown.
Does this benchmark affect the overall score?
This benchmark is preserved as source-native evidence but is not eligible for the current overall score.