Skip to content

Benchmark

IMO 2025 · Raw Points

Higher scores are better.

Models measured
2
Scores recorded
2
Sources
1
IQ weight
No weight published

Score distribution

35
Mean
35
Median
0
Std dev
35
10th percentile
35
90th percentile
35
Min

Leaderboard

#ModelScore
1DeepSeek V3.2 SpecialeSpeciale high-compute reasoning · temperature 1.0 · 128K context · 35/42 · generate-verify-refine · no tools/internet · official contest limits· Reported by DeepSeek· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.35
2MiniMax M3max 42 · test-time scaling ≤10 iterations· Reported by MiniMax Research· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.35

Data credits: BenchBoard by SYuan03 (github.com/SYuan03/benchboard). MIT covers its code and data organization; no results licence is inferred. Numeric relays retain the original stating-source links.