TriggerBench - PM Accuracy: leaderboard
TriggerBench (2026) tests prospective memory in conversation: a user states a constraint early in a dialogue, and a later turn creates a situation where a good assistant should act on it unprompted; a judge model scores whether each reply does. Dialogues are about 2.5K tokens long. The board is judged accuracy in percent over all task variants, including controls where acting would be wrong; higher is better.
Metric: PM accuracy (%) over all 1,265 tasks (positive clean, positive overloaded and negative clean variants); base dialogues of about 2.5K tokens; TriggerBench: 1,265 prospective-memory tasks from 488 dialogue blueprints in five dimensions (state tracking, temporal grounding, logical adherence, attention recovery, safe coding); a constraint is stated early in a dialogue and a later trigger turn calls for acting on it without being asked; GPT-4o (T=0) judges whether the reply fulfils the proactive intent; temperature 0.6; higher is better. Source: arxiv.org. Saturation forecast: Around February 2027. 9 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-5.2 (High) | 83.24 | #176 (GPT-5.2) |
| 2 | GPT-5.2 (Medium) | 81.74 | #176 (GPT-5.2) |
| 3 | Qwen 3 235B A22B 2507 FP8 (Thinking) | 74.86 | #434 |
| 4 | Gemma 3 27B (IT) | 70.59 | #1289 |
| 5 | GPT-5.2 (Non-reasoning) | 69.33 | #176 (GPT-5.2) |
| 6 | GPT-4.1 | 60 | #495 |
| 7 | Qwen 3 235B A22B 2507 Instruct FP8 | 56.21 | #379 |
| 8 | GPT-4o | 51.78 | #741 |
| 9 | Qwen 3 32B | 40.71 | #960 |
No result here: #1 Claude Fable 5.2, #2 Gemini 4 Carbon, #5 Claude Opus 5.5.
Interactive version: theaggregate.ai/benchmark?slug=triggerbench-pm-accuracy · How It Works · Data refreshed daily, snapshot 2026-11-13.