Pi-Bench - Proactivity: leaderboard

Pi-Bench (2026, written as the Greek letter pi) tests proactive personal-assistant agents over 100 long-horizon multi-turn tasks in five professional personas, each starting from an underspecified request with hidden user intents that emerge across sessions; all models run in an agent scaffold adapted from nanobot. This board is proactivity: the percentage of a task hidden intents the agent resolves on its own or elicits with a targeted question, before the simulated user states them, averaged over tasks; higher is better.

Metric: Proactivity (%): share of a task hidden intents the agent resolves without the user stating them, by acting on them or by asking a targeted clarification, averaged over tasks; 100 multi-turn tasks over five professional personas in persistent workspaces, all models under the same agent scaffold adapted from nanobot with thinking enabled and default decoding, a GPT-5.4 simulated user and GPT-5.4 rubric grader, mean of three runs; higher is better. Source: arxiv.org. Saturation forecast: Around February 2027. 9 models tracked. Updated 7 Oct 2026.

Top models

#ModelSettingScoreOverall rank
1GPT-5.4Thinking67#135
2Claude Opus 4.6Thinking65.5#113
3Qwen 3.6 Plus64#238
4GLM-5.158.4#198
5Seed 2.0 Pro58.4#222
6Gemini 3.1 Pro (Preview)57.1#99
7MiniMax-M2.755.6#539
8DeepSeek V3.2Thinking53.3#419
9Kimi K2.543.1#266

No result here: #1 Claude Fable 5.2, #3 Gemini 4 Neon, #4 GPT-6.2 Astra.

Interactive version: theaggregate.ai/benchmark?slug=pi-bench-proactivity · How It Works · Data refreshed daily, snapshot 2026-11-21.