hal-gaia-leaderboard-2026 · post

HAL GAIA Leaderboard (Holistic Agent Leaderboard, Princeton)

Princeton HAL team — hal.cs.princeton.edu

Created: 2026 · Ingested: 2026-09-11

https://hal.cs.princeton.edu/gaia

Princeton's Holistic Agent Leaderboard for GAIA, reporting accuracy per agent-plus-model pairing with the harness named. The rows that matter here: on Claude Sonnet 4.5, HAL's own Generalist Agent scores 74.55% and Hugging Face's Open Deep Research scores 30.91% — the same-model, different-harness comparison that Masood's article and the harness engineering playbook both relay as their headline number.

Referenced by

Claims in this catalog that draw on this source, and whether as support or counterpoint.

supportsA large share of an agent benchmark score is attributable to the harness around the model rather than the model itself: holding the model fixed and changing only guides, sensors, retries and state management is reported to move the same model from 30.9% to 74.6% on GAIA and 30th to 5th on Terminal Bench — so a score reported for "a model" on an agentic benchmark is a score for a model-plus-harness, and comparisons across labs with different harnesses are not comparisons of models.