SemiAnalysis Launches GDPx AI Benchmark
The new benchmark tests frontier AI models on agentic tasks across 11 economically valuable domains, with Anthropic's Claude Opus 5.5 leading at 12.0%.
Updated on Nov. 2, 2026 in Artificial Intelligence
Live Poll
Would you trust an AI system to handle important professional work for you today?
SemiAnalysis unveiled GDPx on November 2, 2026, an AI benchmark evaluating frontier models across 11 domains including chip design, biology, and finance. Anthropic's Claude Opus 5.5 topped the initial leaderboard with a 12.0% overall success rate.
Why it matters
SemiAnalysis argues the first trillion dollars of AI revenue will cluster in sectors where frontier intelligence yields high return on investment, and GDPx is designed to measure progress there. The results indicate that no current model comprehensively solves tasks across all evaluated domains.
GDPx scores on its v1.0 leaderboard are 12.0% for Claude Opus 5.5, 10.9% for Gemini 4 Carbon, and 9.6% for GPT-6 Astra, all well below full task completion. The benchmark excludes the government and insurance sectors, and the source does not explain how the scores were validated beyond the 14 partner startups that built the tasks.
The players
SemiAnalysis
SemiAnalysis is a semiconductor and AI research firm that publishes industry analysis and benchmarks on the economics of compute and frontier models.
Anthropic
Anthropic is an AI safety company that develops the Claude family of large language models.
Google is a multinational technology company whose DeepMind division develops the Gemini family of AI models.
OpenAI
OpenAI is an AI research and deployment company that develops the GPT series of large language models, including ChatGPT.
The details
GDPx selects its domains by where SemiAnalysis believes frontier models will generate the highest return on investment rather than by GDP share or occupational taxonomy. Gemini 4 Carbon led in biology and software engineering, while GPT-6 Astra led in recursive self-improvement and materials science.
Timeline
November 2, 2026: SemiAnalysis unveils the GDPx benchmark.
Every 3 to 6 months: Planned GDPx update cadence, with future updates expected to incorporate feedback and add partners.
The Tech Race
GDPx follows the pattern of task-specific evaluations like SWE-bench, which measure models against defined work rather than general knowledge. By weighting domains by expected return on investment, it treats AI capability as a question of commercial deployment rather than academic performance.
For users choosing among AI tools for professional work, GDPx offers a comparison of task success rates in fields like biology, finance, and software engineering. The low overall scores suggest current models still require substantial human oversight for complex agentic tasks.
The takeaway
Current frontier models solve only a small share of complex agentic tasks across key economic domains, so organizations deploying them should plan for human review. Benchmarks like GDPx, updated every few months, can help track where that gap is closing.
What happens next
SemiAnalysis plans to update the GDPx benchmark every 3 to 6 months, with future versions expected to incorporate feedback and add partners.
Further reading
Track the latest developments in Artificial Intelligence as frontier model benchmarks continue to evolve.
Source note: This article includes information reported by Crypto Briefing.
Live Poll
Would you trust an AI system to handle important professional work for you today?
