Atlas

Benchmarks

← All benchmarks

HMMT Feb 2026

Math · 2026-02-19

This MathArena track evaluates models on the HMMT Feb 2026 problem set. Scores report the percentage of problems answered correctly.

Top models (higher is better)

ModelScore
GPT-5.6 Luna98.5
GPT-5.598.5
GPT-5.497.7
GPT-5.297.0
Kimi K397.0
Opus 4.696.2
Opus 4.895.5
Gemini 3.5 Flash95.5
Gemini 3.1 Pro Preview94.7
Kimi K2.694.7
Opus 4.793.9
DeepSeek-V4-Flash93.9
DeepSeek-V4-Pro93.9
GLM-5.292.4
Inkling-Small90.2
Gemini 3.6 Flash89.4
Gemini 3 Flash Preview89.4
GLM-5.189.4
Qwen3.5 397B A17B87.9
Step 3.7 Flash87.9
Kimi K2.587.1
Gemini 3 Pro Preview86.4
GLM-586.4
Grok 4.1 Fast86.4
Step 3.5 Flash86.4
Inkling86.3
Nemotron 3 Super 120B A12B84.8
DeepSeek-V3.284.1
MiMo-V2.582.6
Qwen3.5 35B A3B81.8
Qwen3.5-27B81.1
Nemotron 3 Ultra 550B A55B78.8
Qwen3 30B A3B Thinking 250778.8
Qwen3.5-4B72.9
Qwen3.5-9B71.2
MiniMax M2.771.2
Haiku 4.566.7
QED-Nano64.4
Gemini 3.5 Flash-Lite63.6
Qwen3 4B Thinking 250753.0
Loading Atlas data…