Indicators / series
hal_reliability.gpt_5_2_medium.accuracy.pt
Source: HAL AI Agent Reliability Tracker (reliability scores by agent) (Princeton University, HAL) · HAL (Princeton); cited with attribution · CSV
HAL: Holistic Agent Leaderboard, AI Agent Reliability Tracker, Princeton University
| id | value | grade | as of date | published date | tier | audited vs reported | extraction method | review status | retrieved at | http status | content hash | supersedes id | url | flags |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 2b6d531d63731873 | 49.9% | grade A, evidence tier 1 | 2025-12-11 | 2026-09-22 | 1 | reported | scrape | approved | 2026-09-22 10:28 | 200 | a0da5842bdd0 | — | source page |
Raw snippets and dispute text
- 2b6d531d63731873 {"label": "GPT-5.2 (medium)", "measure": "accuracy", "provider": "openai", "slug": "gpt-5-2-medium", "x": "2025-12-11", "y": 0.4985314685314685}