Nvidia and ChipAgents launch Build-A-Chip-Bench; top AI agent scores just 16.6%
teortaxesTex · x · 2026-10-15
Nvidia and ChipAgents launched Build-A-Chip-Bench, testing whether AI agents can autonomously design entire chips. The best agent, powered by Anthropic's Claude, scored only 16.6%; 40% of tasks passed functional verification but just 10% met physical requirements for viable chip designs. ChipAgents CEO William Wang says AI has already cut chip development from two years to about one, and expects 4-6 months within a year. Commentator teortaxesTex argues these are strong scores given the models trail the closed frontier by two tiers — chip design can clearly be accelerated without end-to-end autonomy.
Related event: NVIDIA and ChipAgents Launch Chip Design Benchmark; Best Agent Scores 16.6%(3 posts)→
More from Research
- Symmetries and continuous attractors in disordered neural circuits published in Nature Neuroscience — tweetsatpreet · 2026-10-15
- Explainer: why MiniMax's CISPO variant of GRPO preserves the tokens that signal CoT self-correction — andrew_n_carr · 2026-10-15
- Why RL lacks scaling laws: pretraining optimizes loss, open-ended maximization the lightcone — akbirthko · 2026-10-15
- zPocket: open cofolding-feature pocket predictor hits 0.745 AUPRC, beats P2Rank — LotfiSanae · 2026-10-15
- 500-question Gemini study: product named = 78% web search, need described = 7% — MattWags_33 · 2026-10-15
- NCM: new text generation paradigm up to 31x faster than diffusion LMs with better quality — simonguozirui · 2026-10-15