U-MATH - Sequences & Series — leaderboard

Metric: Accuracy (%). Source: huggingface.co. 28 models tracked.

Top models

#ModelScore
1O3 Mini93.51
2DeepSeek R192.86
3O192.86
4Gemini 2.0 Flash (01-21) (Thinking)88.31
5QwQ 32B-Preview84.42
6O1 Mini76.62
7DeepSeek V372.08
8Qwen2.5-Math-72B-Instruct68.83
9QVQ-72B-Preview66.23
10Gemini 1.5 Pro62.99
11Gemini 1.5 Flash60.39
12Athene-V2-Chat57.79
13Qwen 2.5 32B Instruct56.49
14Mistral Large 2 (Nov) Instruct (2411)55.19
15Qwen 2.5 72B Instruct54.55

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