MapSatisfyBench - Explicit Completion: leaderboard

Metric: Explicit-factor completion rate (%): share of the query's stated requirements the final response satisfies; 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: Estimated already saturated. 15 models tracked.

Top models

#ModelScore
1Gemini 3.1 Pro (Preview) (Thinking)91.75
2Claude Opus 4.691.48
3DeepSeek V4 Pro (Non-reasoning)90.88
4Qwen 3.6 Plus (Thinking)90.68
5DeepSeek V3.2 (Non-reasoning)90.19
6GPT-4.189.97
7Claude Sonnet 4.689.83
8Qwen 3.6 Plus (Non-reasoning)89.59
9Qwen 3 235B A22B (Non-reasoning)88.35
10Qwen 3 8B (Non-reasoning)69.97
11Qwen 3 30B A3B (Non-reasoning)60.83

Interactive version: theaggregate.ai/benchmark?slug=mapsatisfybench-explicit-completion · How It Works · Data refreshed daily, snapshot 2026-09-29.