LetTokenBurn
Sign in
Auto57%Harbor: pendingKnowledge & Reasoning

GDPval

GDPval measures model performance on economically valuable, real-world tasks across 44 occupations.

PaperLeaderboard

How to run

ltb · gdpval
$ npm i -g lettokenburn-cli
$ ltb bench pull gdpval

External harness. pip install inspect-evals && inspect eval inspect_evals/gdpval --model <provider/model> https://github.com/UKGovernmentBEIS/inspect_evals

License is link-only: ltb bench pull fetches rows from the original host for local evaluation only and does not write a solution/ directory. Do not redistribute the generated tasks.

Runs happen on your machine with your own subscription; only the score summary is pushed. CLI docs

Scores

11 sourced entries
HarnessSourceConfidenceLink
GPT-5.2-chat49.7N/AIndependentlowN/Aepoch.ai
Claude Opus 4.5 (2025-11-01)45.5N/AIndependentlowN/Aepoch.ai
Claude Opus 4.1 (2025-08-05)43.6N/AIndependentlowN/Aepoch.ai
Claude Sonnet 4.5 (2025-09-29)42.5N/AIndependentlowN/Aepoch.ai
Gemini 3 Pro40.3N/AIndependentlowN/Aepoch.ai
GPT-534.8N/AIndependentlowN/Aepoch.ai
o330.8N/AIndependentlowN/Aepoch.ai
o4-mini25.3N/AIndependentlowN/Aepoch.ai
Gemini 2.5 Pro23.3N/AIndependentlowN/Aepoch.ai
Grok 4 070921.1N/AIndependentlowN/Aepoch.ai
GPT-4o9.9N/AIndependentlowN/Aepoch.ai

Every number links to where it was published. Low-confidence rows are muted and need per-cell verification.

Reproduced by the community

All runs
Loading community runs…

About

GDPval measures model performance on economically valuable, real-world tasks across 44 occupations.