Domain on the map · as Health and bio
Biology: AI benchmarks
- AI benchmarks
- 40
- Shares entries with
- 4 categories
Snapshot 2026-11-12. How categories are built · Where the data comes from
In short
- What are the main biology benchmarks?
- AAV Capsid Packaging Prediction, LifeSciBench, DNA Sequence Design for Transcription Factor Binding, Hard-negative Protein Binding Prediction, Phage–plasmid Co-evolution and SHP2 Protein Function Prediction, among 37 confirmed below.
Only benchmarks use the Biology tag: the RL environment directories file their vendors under other categories. Browse all RL environments.
AI benchmarks 40
-
AAV Capsid Packaging Prediction
Tests health and bio“AAV Capsid Packaging Prediction | Protein design | Can the model predict the viral packaging performance of AAV capsid variants?” source ↗
-
LifeSciBench
Tests health and bio“We introduce LifeSciBench, a benchmark of 750 expert-authored tasks designed to evaluate whether language models can handle realistic life science research work.” source ↗
-
DNA Sequence Design for Transcription Factor Binding
Tests health and bio“Can the model design DNA sequences to maximally bind known transcription factors, with better performance than freely available gradient-based methods?” source ↗
-
Hard-negative Protein Binding Prediction
Tests health and bio“Can the model distinguish ``gain of function'' (binder) protein variants from non-functional (non-binder) proteins, given high quality 3D structure predictions” source ↗
-
Phage–plasmid Co-evolution
Tests health and bio“Can the model predict which amino-acid mutations will arise in a phage–plasmid system at future time points, given earlier observations?” source ↗
-
SHP2 Protein Function Prediction
Tests health and bio“Can the model predict how single-amino-acid changes to SHP2 affect its function?” source ↗
-
BioPipelineBench Verified
Tests health and bio“Anthropic system-card run of BioPipelineBench Verified, measuring biological data-analysis workflow completion.” source ↗
-
Black-box RNA Sequence Design
Tests health and bio“This task measures whether the model can, with minimal prompting and some data access, design RNA sequences in a low-context black-box setting” source ↗
-
DNA Synthesis Screening Evasion
Tests health and bio“tests whether a model can design DNA fragments that assemble into pathogenic viruses while bypassing gene synthesis companies’ DNA screening systems.” source ↗
-
LABBench2 Aggregate
Tests health and bio“LABBench2: An Improved Benchmark for AI Systems Performing Biology Research” source ↗
-
LABBench2 Clinical Trials
Tests health and bio“LABBench2: An Improved Benchmark for AI Systems Performing Biology Research” source ↗
-
LABBench2 Database Questions
Tests health and bio“8.20 Life sciences capabilities” source ↗
-
LABBench2 FigQA
Tests health and bio“8.20 Life sciences capabilities” source ↗
-
LABBench2 Literature Questions
Tests health and bio“8.20 Life sciences capabilities” source ↗
-
LABBench2 Patent Questions
Tests health and bio“LABBench2: An Improved Benchmark for AI Systems Performing Biology Research” source ↗
-
LABBench2 Reading Tables
Tests health and bio“LABBench2: An Improved Benchmark for AI Systems Performing Biology Research” source ↗
-
LABBench2 Supplementary Materials
Tests health and bio“LABBench2: An Improved Benchmark for AI Systems Performing Biology Research” source ↗
-
LatchBio Bioinformatics
Tests health and bio“variant tests the analysis of single-cell RNA sequencing data across 195 problems spanning standard workflows such as labeling cell types” source ↗
-
Long-form Virology Task 1
Tests health and bio“Can agentic systems complete individual tasks related to acquiring, designing, and synthesizing a virus?” source ↗
-
Long-form Virology Task 2
Tests health and bio“Can agentic systems complete individual tasks related to acquiring, designing, and synthesizing a virus?” source ↗
-
Protocol troubleshooting
Tests health and bio“This assessment looks at models’ ability to detect and fix errors in molecular biology protocols” source ↗
-
Structural Biology Open-Ended
Tests health and bio“8.16.4 Structural biology, open-ended” source ↗
-
VCT Virology Capabilities Test
Tests health and bio“a large language model (LLM) benchmark that measures the capability to troubleshoot complex virology laboratory protocols” source ↗
-
ViroBench
Tests health and bio“ViroBench, the first comprehensive and large-scale benchmark specifically designed for NFMs in viral settings” source ↗
-
VariantBench
Tests health and bio“We present VariantBench, a verifiable benchmark of 118 evaluations that tests agents on their ability to conduct variant discovery and interpretation.” source ↗
-
PlantMarkerBench
Tests health and bio“We introduce PlantMarkerBench, a multi-species benchmark for evaluating literature-grounded plant marker evidence interpretation from full-text biological papers.” source ↗
-
SciHorizon-GENE
Tests health and bio“we introduce SciHorizon-GENE, a large-scale gene-centric benchmark constructed from authoritative biological databases.” source ↗
-
BioCoder
Tests health and bio“They contain domain-specific knowledge of bioinformatics, beyond just general coding capability.” source ↗
-
BioSecBench-Function
Tests health and bio“BioSecBench-Function measures whether AI agents can turn real experimental data, biophysical assay readouts, structures, or sequences, into a correct functional or biosecurity-relevant conclusion” source ↗
-
BioSecBench-Surveillance
Tests health and bio“A verifiable benchmark for AI agents on pathogen genomic surveillance” source ↗
-
Multimodal Troubleshooting Virology
Tests health and bio“How well can models perform on virology questions testing protocol troubleshooting?” source ↗
-
ProtocolQA Open-Ended
Tests health and bio“How well can models perform on open-ended questions testing protocol troubleshooting?” source ↗
-
Tacit Knowledge and Troubleshooting
Tests health and bio“Can models answer as well as experts on difficult tacit knowledge and troubleshooting questions?” source ↗
-
TroubleshootingBench
Tests health and bio“To evaluate models’ ability to identify and correct real-world experimental errors in biological protocols, we built a short-answer troubleshooting dataset from expert-written wet lab procedures.” source ↗
-
scBench-Long
Tests health and bio“We introduce scBench-Long, a benchmark for long-horizon single-cell biology in which agents must recover scientific conclusions from raw or near-raw data” source ↗
-
SpatialBench-Long
Tests health and bio“We introduce SpatialBench-Long, a benchmark for long-horizon spatial biology in which agents must recover biological claims from raw or near-raw data” source ↗
-
SynBio-Reason
Tests health and bio“SynBio-Reason, a benchmark of 4,753 circuits spanning six canonical circuit types and nine tasks from code repair to de novo design” source ↗
3 not yet confirmedListed in the catalog; no line from its own site or paper confirms it yet.
Data 16
- AfterQuery
- Align AI
- Anthromind
-
Aptura
Sells data“Physicians, attorneys, and investment analysts with years of high-stakes experience, supported by a platform built for structured, high-quality data collection at scale.” source ↗
Covers health and bio“I review oncology cases from home to help train AI models that support cancer diagnosis.” source ↗
- Handshake
- Mercor
- micro1
- ReasonCore
- Refresh
- Rise Data Labs
- Scale AI
-
Sepal AI
Sells data“Frontier AI needs frontier experts. Work with Sepal AI to fuel safer and smarter products and unlock new possibilities for humanity.” source ↗
Covers health and bio“We are developing a groundbreaking benchmark to rigorously evaluate the ability of AI systems to reason about complex biological systems and phenomena.” source ↗
- Surge AI
-
Theta
Covers health and bio“Analyze Patient Labs” source ↗
- Turing
- Vetto AI
Next: pick another domain. Tags come from the RL Research daily scrape, canonicalized nightly.