BenchLeader

USAMO 2026

USA Mathematical Olympiad 2026, proof-graded by MathArena. Shown for reference; too few models, too close to the ceiling.

As of 21 Oct 2026, GPT-5.5 leads USAMO 2026 on BenchLeader with 98.2%, ahead of GPT-5.4 at 95.2%, across 9 model configurations with a published result.

MathArena marks 3 of these 9 results as possibly contaminated: the model was evaluated after the questions were public, so they may already be in its training data. Those scores are in the table below; this board does not feed the index.

Published by
MathArena
Category
Maths
Index weight
Reference only
Models
9
Data as of
21 Oct 2026

MIT — eth-sri/matharena.

9 of 9
#
1GPT-5.5xhigh98.2%66.1possible
2GPT-5.4xhigh95.2%64.8–
3Gemini 3.1 Pro74.4%60.7–
4DeepSeek V4 Promax60.7%60.8possible
5Kimi K2.6thinking51.2%59.4possible
6Claude Opus 4.6high47.0%62.8–
7Step 3.5 Flash44.6%53.8–
8Qwen3.5 397B A17B36.3%51.6–
9GLM-535.1%55.0–

More maths benchmarks · All LLM benchmarks

Cite as: BenchLeader, “USAMO 2026 leaderboard”, https://www.benchleader.com/benchmarks/matharena_usamo_2026, data as of 21 Oct 2026.

USAMO 2026: questions

What does USAMO 2026 measure?
USA Mathematical Olympiad 2026, proof-graded by MathArena. Shown for reference; too few models, too close to the ceiling. Scores are reported in percent of tasks solved; higher is better.
Which AI model leads USAMO 2026?
GPT-5.5 leads USAMO 2026 with 98.2% as of 21 Oct 2026, ahead of GPT-5.4 at 95.2%.
How many models have USAMO 2026 results?
9 model configurations have a USAMO 2026 result on BenchLeader, all taken from MathArena.
Who runs USAMO 2026 and how often is it updated?
USAMO 2026 is published by MathArena. BenchLeader re-reads the published results once a day and records the date each result was published.
Does USAMO 2026 count toward the BenchLeader Index?
No. USAMO 2026 is shown for reference but left out of the composite index.