FrontierCS 2 benchmark debuts: Claude Opus 5.5 leads at 58.3%, Gemini 4 Carbon at 54.2%
eliebakouch · x · 2026-11-16
FrontierCS 2 is a new benchmark for AI computer-science R&D, comprising 48 tasks with 5 hours each and a uniform pass@1 metric.
- Claude Opus 5.5: 58.3% (28/48), ranking first
- Gemini 4 Carbon: 54.2% (26/48), ranking second
It targets research-grade CS problems harder than standard coding benchmarks, offering a new yardstick for frontier-model research capability.
More from Models
- Claude's formulaic creative writing sparks Reddit debate on editing AI output — GIMBruhBoobies · 2026-11-16
- User claims GPT-6 'Astra' can already control robots and fine-tuned an open model's weights — xmercury_one · 2026-11-16
- Swarm orchestration runs 12 parallel agents to build a Blender 3D model, claims tester — venturetwins · 2026-11-16
- Gemini Robotics 2.5 debuts with 96% pick-and-pack success on Mercedes-Benz lines — dl_weekly · 2026-11-16
- Google solicits feedback on 'Gemini 4 Argon, Carbon, Hydrogen' in apparent leak — deliprao · 2026-11-16
- James Douma: Grok is currently the best tool for understanding what's happening in the world — jamesdouma · 2026-11-16