llmboard.ai
Benchmarks
CompareRankings
llmboard.ai
Benchmarks
CompareRankings
HomeBenchmarksmathHMMT 2025

math benchmark

HMMT 2025

Harvard-MIT Mathematics Tournament 2025 - A prestigious student-organized mathematics competition for high school students featuring two tournaments (November 2025 at MIT and February 2026 at Harvard) with individual tests, team rounds, and guts rounds

Updated Aug 7, 2026

Published models33
Registry coverage33
MetricScore
EvidenceB

On this page

  • Ranking
  • Distribution
  • Highlights
  • About
  • FAQ

HMMT 2025 leaderboard

Sorted by the source-provided rank. Higher score is better according to the registry.

33 rows
Columns

Show columns

1OPGPT-5.2 ProOpenAI100.0%100.0%33CAug 7, 2026
2OPGPT-5.2OpenAI99.4%96.9%33CAug 7, 2026
3DEDeepSeek-V3.2-SpecialeDeepSeek99.2%93.8%33CAug 7, 2026
4MAKimi K2-Thinking-0905Moonshot AI97.5%90.6%33CAug 7, 2026
5ACQwen3.6 PlusAlibaba Cloud / Qwen Team96.7%87.5%33CAug 7, 2026
6MAKimi K2.5Moonshot AI95.4%84.4%33CAug 7, 2026
7ACQwen3.5-397B-A17BAlibaba Cloud / Qwen Team94.8%81.3%33CAug 7, 2026
8NVNemotron 3 Super (120B A12B)NVIDIA94.7%78.1%33CAug 7, 2026
9ZAGLM-5.2Zhipu AI94.4%75.0%33CAug 7, 2026
10ZAGLM-5.1Zhipu AI94.0%71.9%33CAug 7, 2026
11ACQwen3.6-27BAlibaba Cloud / Qwen Team93.8%68.8%33CAug 7, 2026
12OPGPT-5OpenAI93.3%65.6%33CAug 7, 2026
13XAGrok 4 FastxAI93.3%62.5%33CAug 7, 2026
14ACQwen3.5-27BAlibaba Cloud / Qwen Team92.0%59.4%33CAug 7, 2026
15ACQwen3.5-122B-A10BAlibaba Cloud / Qwen Team91.4%56.3%33CAug 7, 2026
16ACQwen3.6-35B-A3BAlibaba Cloud / Qwen Team90.7%53.1%33CAug 7, 2026
17DEDeepSeek-V3.2 (Thinking)DeepSeek90.2%50.0%33CAug 7, 2026
18DEDeepSeek-V3.2DeepSeek90.2%46.9%33CAug 7, 2026
19ACQwen3.5-35B-A3BAlibaba Cloud / Qwen Team89.0%43.8%33CAug 7, 2026
20OPGPT-5 miniOpenAI87.8%40.6%33CAug 7, 2026
21SASarvam-105BSarvam AI85.8%37.5%33CAug 7, 2026
22XIMiMo-V2-FlashXiaomi84.4%34.4%33CAug 7, 2026
23DEDeepSeek-V3.2-ExpDeepSeek83.6%31.3%33CAug 7, 2026
24ACQwen3.5-9BAlibaba Cloud / Qwen Team83.2%28.1%33CAug 7, 2026
25DEDeepSeek-R1-0528DeepSeek79.4%25.0%33CAug 7, 2026
26OPGPT-5 nanoOpenAI75.6%21.9%33CAug 7, 2026
27ACQwen3.5-4BAlibaba Cloud / Qwen Team74.0%18.8%33CAug 7, 2026
28SASarvam-30BSarvam AI73.3%15.6%33CAug 7, 2026
29MAKimi K2 InstructMoonshot AI38.8%12.5%33CAug 7, 2026
30MAKimi K2-Instruct-0905Moonshot AI38.8%9.4%33CAug 7, 2026
31OPGPT-4.1 miniOpenAI35.0%6.3%33CAug 7, 2026
32DEDeepSeek-V3.1DeepSeek33.5%3.1%33CAug 7, 2026
33OPGPT-4.1OpenAI28.9%0.0%33CAug 7, 2026

Score distribution

Top published rows on the benchmark's original scale.

HMMT 2025

HMMT 2025 highlights

The top published results on this benchmark's own scale.

Rank #1GPT-5.2 Pro100.0%Rank #2GPT-5.299.4%Rank #3DeepSeek-V3.2-Speciale99.2%Rank #4Kimi K2-Thinking-090597.5%

What is HMMT 2025?

Definition and scoring fields from the benchmark registry.

Harvard-MIT Mathematics Tournament 2025 - A prestigious student-organized mathematics competition for high school students featuring two tournaments (November 2025 at MIT and February 2026 at Harvard) with individual tests, team rounds, and guts rounds

Scores are shown in ratio. The current registry marks this benchmark as not independently verified with evidence level B.

Family
HMMT 2025
Modality
text
Primary category
math
Score direction
higher
LLMBoard eligible
Yes
Evaluation key
hmmt-2025|llm-stats-current

Source-native results are preserved. Eligibility for the overall LLMBoard score is a separate policy decision.

FAQ

Common questions about HMMT 2025.

Which model scores highest on HMMT 2025?

GPT-5.2 Pro is currently ranked first with 100.0%.

What does HMMT 2025 measure?

Harvard-MIT Mathematics Tournament 2025 - A prestigious student-organized mathematics competition for high school students featuring two tournaments (November 2025 at MIT and February 2026 at Harvard) with individual tests, team rounds, and guts rounds

Is a higher score better?

Yes. Higher values rank better for this benchmark.

How many models are compared?

33 unique published model results are currently shown.

Does this benchmark affect the overall score?

This benchmark is marked as eligible for the current LLMBoard capability methodology.

Rankings

OverallCodingText ArenaPricing

Modalities

Image GenerationVideo GenerationSpeech-to-TextEmbeddings

Benchmarks

All BenchmarksReasoningMathCoding

Vendors

All VendorsOpenAIAnthropicGoogle
llmboard.aiCopyright 2026 llmboard.ai