Atlas

Benchmarks

← All benchmarks

Kangaroo 2025 9-10

Math · 2025-10-20

This MathArena track evaluates models on the Kangaroo 2025 9-10 problem set. Scores report the percentage of problems answered correctly.

Top models (higher is better)

ModelScore
Gemini 3.1 Pro Preview100.0
Gemini 3.5 Flash100.0
Gemini 3.6 Flash100.0
GPT-5.2100.0
GPT-5.5100.0
Kimi K3100.0
GPT-5.499.2
Gemini 3 Flash Preview98.3
GPT-5 Mini97.5
Gemini 3 Pro Preview96.7
Gemini 2.5 Pro95.8
Kimi K2.595.8
Sonnet 4.595.0
GPT-592.5
Opus 4.691.7
Opus 4.891.7
GPT-5.190.8
Qwen3-VL-235B-A22B-Instruct89.2
Grok 4.1 Fast87.5
Grok 4 Fast87.5
Grok 485.8
GLM-4.5V78.3
Loading Atlas data…