LatchBio launches OmicsBench v1, a 784-task meta-benchmark for omics AI
kenbwork · x · 2026-10-20
The LatchBio team (kenbwork et al.) introduces OmicsBench v1, a meta-benchmark meant as a high-level barometer of model progress on molecular ("omics") data analysis.
- Composition: 784 tasks drawn from eight benchmarks — SpatialBench, ScBench, EpiBench, VariantBench, BioSecBench-Surveillance, BioSecBench-Function, SpatialBench-Long, ScBench-Long
- Motivation: after releasing many domain-specific agent benchmarks covering bioinformatics, therapeutics and biosecurity, the team argues enough data types have been measured to warrant a meta-benchmark
- The accompanying figure compares pass rates on short- and long-horizon tasks across frontier models (Claude, GPT, Gemini, Grok, Kimi, DeepSeek families) over 2026 release dates
- Team member Chris published a companion essay, with more analysis to follow
More from Research
- Single neuron steering makes Qwen3-8B insist it's an owl — Sauers_ · 2026-10-20
- Q-CTRL: over 99.5% of qubits sit idle in one fault-tolerant Shor's algorithm blueprint — MJBiercuk · 2026-10-20
- No Siegel zeros proved in just 3 pages, echoing the polynomial method's history — soumitrashukla9 · 2026-10-20
- Logarithmic returns come from opportunities spread across orders of magnitude of cost — DimitrisPapail · 2026-10-20
- Emergent curriculum without labels: prompt weights diverge 1.6x vs 0.75x over training — shangbinfeng · 2026-10-20
- Stackelberg Alignment turns LLM disagreement into a training curriculum via bandit leader — shangbinfeng · 2026-10-20