Atlas

Benchmarks

← All benchmarks

Kangaroo 2025 5-6

Math · 2025-10-20

This MathArena track evaluates models on the Kangaroo 2025 5-6 problem set. Scores report the percentage of problems answered correctly.

Top models (higher is better)

ModelScore
Gemini 3.5 Flash90.0
GPT-5.590.0
Gemini 3.6 Flash88.3
Gemini 3.1 Pro Preview86.7
GPT-5.483.3
Kimi K381.7
GPT-5.280.0
Gemini 3 Flash Preview77.5
Gemini 3 Pro Preview76.7
Qwen3.5 397B A17B74.9
GPT-5 Mini70.8
Opus 4.870.0
Sonnet 4.568.3
Kimi K2.567.5
Gemini 2.5 Pro66.7
Grok 4.1 Fast65.8
GPT-565.0
Grok 463.3
GLM-4.5V62.5
GPT-5.161.7
Grok 4 Fast61.7
Qwen3-VL-235B-A22B-Instruct60.8
Opus 4.658.3
Loading Atlas data…