OncologyBench: 25 Multistep Cancer Evaluations, Best Model Scores Only 40%
kenbwork · x · 2026-10-14
kenbwork's team released OncologyBench, a verifiable benchmark of 25 multistep oncology evaluations (built on 92 component tasks) spanning translational research, detection, diagnosis, treatment and research priorities. Tasks are deterministically graded against the biological result of the original analysis, refusals and timeouts count in the denominator, and scores are reported per model+harness.
Scores are bleak: Claude Opus 5 + Pi leads at just 40.0%, Claude Opus 5.5 + Claude Code at 37.3%; Kimi K3 24.0%, GPT-6 Astra + Pi 21.3%, Grok 4.6 + Pi 18.7%. Leaderboard, paper, blog and GitHub are all public.
Related event: OncologyBench Released: Top Models Pass Only 29-40% of Oncology Decisions(7 posts)→
More from Models
- Google's gemini-4-demo-vanilla-3D repo surfaces, hinting at Gemini 4 3D demos — fofrAI · 2026-10-14
- Gemini 4 Argon goes live on Vercel AI Gateway at 50% off as rauchg hails coding agents as AI's killer app — AccBalanced · 2026-10-14
- DIY CrusaderBench ranks Gemini 4 Argon #1 at Stronghold strategy, beating Opus 5.5 — Hesamation · 2026-10-14
- Meta's Muse Wonder ships without a single cost-vs-score plot, researchers cry foul — AccBalanced · 2026-10-14
- Google offers free Antigravity access to researchers, seizing on OpenAI's mathematician backlash — altesc_create · 2026-10-14
- Designer: new model is highly capable at design but defaults to generic 'AI slop' look unsteered — zacharynado · 2026-10-14