GDPx: SemiAnalysis's frontier benchmark across the 11 most valuable domains
rohanpaul_ai · x · 2026-11-02
SemiAnalysis (Dylan Patel et al.) published GDPx, a benchmark measuring frontier agentic capability across 11 high-ROI domains: chip design, cyber, bio, finance, legal, quant trading, RSI, hardware design, materials science, alignment, and software engineering. Thesis: the first trillion dollars of AI revenue will concentrate in sectors deploying frontier intelligence at high ROI, so GDP-share-based benchmarks are the wrong target. Top score is a mere 12% (Claude Opus 5.5); excluding cyber/bio/alignment, where safety classifiers intervene, GPT-6 Astra rises to 11.8% and overtakes Opus. OpenAI and Anthropic dominate; Meta, xAI, and open-weight models trail. Finance and law zero out every model under pass-fail grading, and models worsened a working trading strategy in 70 of 111 attempts. Unlike GDPval, agents must explore real workspaces (e.g., a 3,983-file acquisition replica) rather than receive context in prompts.
Related event: SemiAnalysis Launches GDPx Frontier Benchmark; Best Model Scores Just 12%(10 posts)→
More from Models
- ICML paper: finite context windows impose impossibility results for adversarial regret — burny_tech · 2026-11-02
- Fable 5.5 creates 66-second animated short entirely with self-written code, no image models — oran_ge · 2026-11-02
- SenseTime open-sources SenseNova-Loop-9B: weight-shared 56-layer loop, 256K context, Apache 2.0 — AdinaYakup · 2026-11-02
- User says Gemini excels at long-form fanfic, producing stories up to 29 minutes of reading — Ok-Jello3739 · 2026-11-02
- Sam Altman calls tokens 'a kind of fake unit' that doesn't measure business value — rohanpaul_ai · 2026-11-02
- 'You Can Feel the Gradient Slopes': Dev Contrasts Benchmaxxed RL Models vs General Ones — theteknosaur · 2026-11-02