Share your thoughts, 1 month free Claude Pro on usSee more

Home/Benchmarks

Agent Confidence Calibration on GAIA (val)

0.382ECE

LastStep-TP

Updated 1mo ago

Evaluation Results

Method	Links
LastStep-TP 2026.01		0.382	0.375	0.607

SOTA Paper

LastStep-TP

Agentic Confidence Calibration

Dataset

GAIA

Follow for update

@wizwand_team Discord

Related Benchmarks

Confidence Calibration on SimpleQA (test)Confidence Calibration on GPQA (test)Confidence Calibration on HLE (test)

© 2026 wizwand

Blog Contact Changelog Swarm

Privacy Policy Terms of Service FAQs Swarm Docs