Inference
EN

Harvard-MIT Mathematics Tournament February 2025

ModelsModels with a result.15
Top resultBest result on this benchmark.97.5%GLM-5
Top-3 spreadPoints from first to third.0.8 pts
YearYear of release.—

Result and price

10 of 15 results

Results

15 results
#ModelResultThe score from the source.
01GLM-597.5%
02GLM-4.797.1%
03Qwen3.6 Plus96.7%
04Kimi K2.5 (Reasoning)95.4%
04Kimi K2.595.4%
06Qwen3.5 397B94.8%
07Qwen3.6-27B93.8%
08Nemotron 3 Super 100B93.7%
09Claude Opus 4.592.9%
10Qwen3.6-35B-A3B90.7%
11Granite 4.2 30B89.2%
12Sarvam 105B85.8%
13Granite 4.2 8B78.3%
14Sarvam 30B73.3%
15Granite 4.2 3B66.7%