FrontierMath Tier 4 (v2) Leaderboard

51 of 51 models

AI

AI co-mathematician

Claude Fable 5 (max)

Claude Opus 4.1 (32k thinking)

Claude Opus 4.5 (32k thinking)

Claude Opus 4.6 (max)

Claude Opus 4.7 (max)

Claude Opus 4.8 (max)

Claude Opus 5 (max)

Claude Sonnet 4.5 (32k thinking)

Claude Sonnet 5 (max)

DeepSeek v4 (max)

DeepSeek V4 Flash 0731 (max)

DeepSeek V4 Pro 0813 (max)

Gemini 2.5 Pro

Gemini 3 Flash Preview

Gemini 3.1 Pro Preview

Gemini 3.5 Flash (high)

Gemini 3.5 Flash-Lite (high)

Gemini 3.6 Flash (high)

Gemini 3.7 Flash (high)

GLM-5.2 (max)

GPT-5 (high)

GPT-5 mini (high)

GPT-5 nano (high)

GPT-5 Pro (high)

GPT-5.2 (xhigh)

GPT-5.2 Pro (xhigh)

GPT-5.4 (xhigh)

GPT-5.4 mini (xhigh)

GPT-5.4 nano (high)

GPT-5.4 Pro (xhigh)

GPT-5.5 (xhigh)

GPT-5.5 Instant

GPT-5.5 Pro (xhigh)

GPT-5.6 Luna (max)

GPT-5.6 Sol (max)

GPT-5.6 Sol (pro, max)

GPT-5.6 Terra (max)

Grok 4.20

Grok 4.3 Beta

Grok 4.5 (high)

Grok 4.6 (xhigh)

In

Inkling (xhigh)

In

Inkling Small (xhigh)

Kimi K2.6

Kimi K2.7 Code

Kimi K3 (Max)

o3-mini (high)

o4-mini (high)

Qwen3.7 Max

Qwen3.8 Max (xhigh)

51 of 51 models

Higher is better

List bars visualize relative performance. Plot bars use published 0–100 scores, shrink to a 1.5rem minimum, then scroll horizontally.