ReliableMath - Precision — leaderboard

Metric: Score (%). Source: huggingface.co. 18 models tracked.

Top models

#ModelScore
1DeepSeek R164.2
2Qwen 3 235B A22B62.1
3doubao-seed-1.6-thinking-25061559.4
4Qwen 3 14B57.3
5DeepSeek R1 052856.9
6DeepSeek R1 Distill Qwen 32B55.1
7DeepSeek R1 Distill Qwen 14B54.7
8Qwen 3 32B54.5
9DeepSeek V352.1
10Claude Sonnet 4 (Thinking)52
11Gemini 2.5 Flash (Preview 04-17)51.8
12Gemini 2.5 Flash (Preview 04-17) (Thinking)50.8
13O3 Mini (2025-01-31)50.4
14Claude Sonnet 4 (20250514)47.3
15GPT-4o (2024-08-06)39.7

Interactive version: aibenchmarks.dev/benchmark?slug=reliablemath-precision · How the rankings work · Data refreshed daily, snapshot 2026-07-08.