MapSatisfyBench - Implicit Intent Satisfaction: leaderboard
Metric: Implicit-intent satisfaction rate (%): evidence-weighted satisfaction of the unstated decision factors (hard constraints binary, soft preferences graded); underspecified map-service requests rebuilt from real anonymized user behavior chains, answered by a ReAct agent with 22 deterministic replayed map tools and a simulated user; GPT-5.3 simulates the user and judges (majority of three judgments), temperature 1; higher is better. Source: arxiv.org. Saturation forecast: Around December 2026. 15 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Claude Opus 4.6 | 71.7 |
| 2 | DeepSeek V4 Pro (Non-reasoning) | 70.17 |
| 3 | Qwen 3.6 Plus (Thinking) | 67.83 |
| 4 | Claude Sonnet 4.6 | 67.25 |
| 5 | DeepSeek V3.2 (Non-reasoning) | 66.1 |
| 6 | Qwen 3.6 Plus (Non-reasoning) | 65.85 |
| 7 | Gemini 3.1 Pro (Preview) (Thinking) | 64.69 |
| 8 | Qwen 3 235B A22B (Non-reasoning) | 62.37 |
| 9 | GPT-4.1 | 61.04 |
| 10 | Qwen 3 8B (Non-reasoning) | 37.16 |
| 11 | Qwen 3 30B A3B (Non-reasoning) | 36.35 |
Interactive version: theaggregate.ai/benchmark?slug=mapsatisfybench-implicit-intent-satisfaction · How It Works · Data refreshed daily, snapshot 2026-09-29.