MapSatisfyBench - Tool Selection: leaderboard
MapSatisfyBench (2026) gives map-service agents informal, underspecified requests rebuilt from real user behavior and checks whether they recover the unstated needs through 22 replayed map tools and a simulated user, with GPT-5.3 as judge. This board is the set overlap in percent between the tools the agent validly called and the required tools, penalizing missing and redundant calls; higher is better.
Metric: Tool-selection score (%): set overlap between the valid tools the agent called and the annotated required tools, penalizing missing and redundant calls; 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 2029. 15 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Claude Opus 4.6 | 49.42 | #102 |
| 2 | Claude Sonnet 4.6 | 49.09 | #161 |
| 3 | GPT-4.1 | 46.96 | #495 |
| 4 | Gemini 3.1 Pro (Preview) (Thinking) | 45.6 | #94 (Gemini 3.1 Pro (Preview)) |
| 5 | Qwen 3.6 Plus (Thinking) | 45.36 | #222 (Qwen 3.6 Plus) |
| 6 | Qwen 3.6 Plus (Non-reasoning) | 44.93 | #222 (Qwen 3.6 Plus) |
| 7 | DeepSeek V4 Pro (Non-reasoning) | 43.82 | #188 (DeepSeek V4 Pro) |
| 8 | DeepSeek V3.2 (Non-reasoning) | 43.79 | #406 (DeepSeek V3.2) |
| 9 | Qwen 3 235B A22B (Non-reasoning) | 42.36 | #623 (Qwen 3 235B A22B) |
| 10 | Qwen 3 8B (Non-reasoning) | 33.49 | #1830 (Qwen 3 8B) |
| 11 | Qwen 3 30B A3B (Non-reasoning) | 28.59 | #1122 (Qwen 3 30B A3B) |
No result here: #1 Claude Fable 5.2, #2 Gemini 4 Carbon, #5 Claude Opus 5.5.
Interactive version: theaggregate.ai/benchmark?slug=mapsatisfybench-tool-selection · How It Works · Data refreshed daily, snapshot 2026-11-13.