No AI agent tops 12% on SemiAnalysis' real-world GDPx benchmark
The Decoder · rss · 2026-11-06
SemiAnalysis' new GDPx benchmark puts AI agents through simulated knowledge work spanning chip design to quant trading. No model breaks 12%, with Claude Opus 5.5 narrowly leading Gemini 4 Carbon. Critics note the small number of tasks per domain and uneven testing conditions.
More from Models
- Blogger's head-to-head test: Gemini 4 Argon beats Opus 5.5 at dissecting viral posts — huangyun_122 · 2026-11-06
- Accentus filters speech data: rejects 2% of Americans, 93% of non-American accents — HeyAmit_ · 2026-11-06
- Reasoning hurts pointwise reranking: EMNLP paper shows Rank1 underperforms — lintool · 2026-11-06
- OpenAI's cyber model found 948 Linux vulns and 111 Firefox RCEs, says Greg Brockman — gdb · 2026-11-06
- GPT-6.1 Sol halves surgical positioning error after 3 mouse brain surgeries — 141_1337 · 2026-11-06
- Ripple-65B Preview hits 20.5 tok/s for code but only 7.6 for prose on a single DGX Spark — DZhang50 · 2026-11-06