Gemini 4 tops RareBench at 57.4% and LabBench at 52.2%, dethroning Claude Opus 5.5
danielmckinn0n · x · 2026-10-14
Developer danielmckinn0n reports that Google DeepMind's Gemini 4 takes the crown on GamowLabs' RareBench and LabBench science benchmarks, beating previous champion Claude Opus 5.5 by a healthy margin. Charts show Gemini 4 Argon scoring 57.4% on RareBench complete causal recall (79/122 cases) vs Opus 5.5's 51.6% and Fable 5's 42.6%, with GPT-6.1 Sol at just 32.0%; on LabBench mean task score, Gemini 4 Argon hits 52.2% vs GPT-6 Astra's 40.8% and Opus 5.5's 40.5%. The poster says his team has moved essentially all biology and genomics research work to Gemini 4, hoping it can help diagnose new pediatric rare-disease cases.
Related event: Gemini 4 Tops RareBench and LabBench, Dethroning Opus 5.5(2 posts)→
More from Models
- Unverified screenshots surface of a supposed Gemini 4 Argon model — ChrisUniverse · 2026-10-14
- Third-party testing suggests Mistral 4 is heavily benchmaxxed, roughly a year behind frontier — teortaxesTex · 2026-10-14
- Dev says open source lets his Sora-style video business turn a profit where OpenAI can't — ctjlewis · 2026-10-14
- Gemini 4 Argon ships in GitHub Copilot, beating prior Gemini models on every coding benchmark — rseroter · 2026-10-14
- Gemini 4 Argon appears in the app, then refuses to talk about its own release — Cyrone007 · 2026-10-14
- AnythingLLM ditches Ollama for llama.cpp, with up to 2x faster inference — solyarisoftware · 2026-10-14