HMMT25

Name: HMMT25 Leaderboard — AI Model Scores
Creator: LLM Stats
License: https://llm-stats.com/legal/terms-of-service

Paper

Progress Over Time

Interactive timeline showing model performance evolution on HMMT25

State-of-the-art frontier

Open

Proprietary

HMMT25 Leaderboard

28 models

			Context	Cost
1	Grok-4 Heavy xAI	—	—	—
2	Qwen3.6 Plus Alibaba Cloud / Qwen Team	—	1.0M	$0.50 / $3.00
3	Qwen3.5-397B-A17B Alibaba Cloud / Qwen Team	397B	262K	$0.45 / $3.00
4	Qwen3.6-27B Alibaba Cloud / Qwen Team	28B	262K	$0.32 / $3.20
5	Qwen3.5-122B-A10B Alibaba Cloud / Qwen Team	122B	262K	$0.29 / $2.40
6	Grok-4 xAI	—	—	—
7	Qwen3.5-27B Alibaba Cloud / Qwen Team	27B	262K	$0.26 / $2.60
8	Qwen3.5-35B-A3B Alibaba Cloud / Qwen Team	35B	262K	$0.14 / $1.00
9	IBM Granite 4.2 30B IBM	30B	131K	$0.16 / $0.65
10	Qwen3.6-35B-A3B Alibaba Cloud / Qwen Team	35B	262K	$0.10 / $0.95
11	Sarvam-105B Sarvam AI	105B	—	—
12	Qwen3-235B-A22B-Thinking-2507 Alibaba Cloud / Qwen Team	235B	—	—
13	Qwen3.5-9B Alibaba Cloud / Qwen Team	9B	262K	$0.10 / $0.15
14	IBM Granite 4.2 8B IBM	8B	131K	$0.06 / $0.25
15	Qwen3 VL 235B A22B Thinking Alibaba Cloud / Qwen Team	236B	—	—
16	Qwen3.5-4B Alibaba Cloud / Qwen Team	4B	—	—
17	Sarvam-30B Sarvam AI	30B	—	—
18	Qwen3-Next-80B-A3B-Thinking Alibaba Cloud / Qwen Team	80B	—	—
19	Qwen3 VL 30B A3B Thinking Alibaba Cloud / Qwen Team	31B	—	—
20	IBM Granite 4.2 3B IBM	3B	131K	$0.03 / $0.12
21	Qwen3 VL 8B Thinking Alibaba Cloud / Qwen Team	9B	—	—
22	Qwen3 VL 235B A22B Instruct Alibaba Cloud / Qwen Team	236B	262K	$0.20 / $0.88
23	Qwen3-235B-A22B-Instruct-2507 Alibaba Cloud / Qwen Team	235B	262K	$0.09 / $0.55
24	Qwen3-Next-80B-A3B-Instruct Alibaba Cloud / Qwen Team	80B	262K	$0.09 / $1.10
25	Qwen3 VL 4B Thinking Alibaba Cloud / Qwen Team	4B	—	—
26	Qwen3 VL 30B A3B Instruct Alibaba Cloud / Qwen Team	31B	262K	$0.15 / $0.60
27	Qwen3 VL 8B Instruct Alibaba Cloud / Qwen Team	9B	—	—
28	Qwen3 VL 4B Instruct Alibaba Cloud / Qwen Team	4B	—	—

Notice missing or incorrect data?

About this benchmark

What is HMMT25?

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

HMMT25 is a text benchmark evaluating models on math tasks. LLM Stats tracks 28 models on this benchmark, scored on a 0–1 scale. The current average is 0.7, with the leader at 1.0.

Compare leaders on the best AI for math leaderboards.

Current leaders

Grok-4 Heavy from xAI currently leads the HMMT25 leaderboard with a score of 0.967 across 28 evaluated AI models.

Grok-4 HeavyxAI96.7%

Qwen3.6 PlusAlibaba Cloud / Qwen Team94.6%

Qwen3.5-397B-A17BAlibaba Cloud / Qwen Team92.7%

FAQ

Common questions about the HMMT25 benchmark and leaderboard.

What is the HMMT25 benchmark?

What is the HMMT25 leaderboard?

The HMMT25 leaderboard ranks 28 AI models based on their performance on this benchmark. Currently, Grok-4 Heavy by xAI leads with a score of 0.967. The average score across all models is 0.741.

What is the highest HMMT25 score?

The highest HMMT25 score is 0.967, achieved by Grok-4 Heavy from xAI.

How many models are evaluated on HMMT25?

28 models have been evaluated on the HMMT25 benchmark, with 0 verified results and 28 self-reported results.

Where can I find the HMMT25 paper?

The HMMT25 paper is available at http://web.mit.edu/HMMT/www/. The paper details the methodology, dataset construction, and evaluation criteria.

What categories does HMMT25 cover?

HMMT25 is categorized under math. The benchmark evaluates text models.

What is the best open-source model on HMMT25?

Qwen3.5-397B-A17B by Alibaba Cloud / Qwen Team is the top-ranked open-source model on HMMT25, with a score of 0.927 (rank #3).

Which model offers the best value on HMMT25?

Among models scoring within 10% of the leader, Qwen3.6-35B-A3B from Alibaba Cloud / Qwen Team is the cheapest, at $0.10 per million input tokens with a score of 0.891.

How recent are the HMMT25 leaderboard results?

The HMMT25 leaderboard was last updated in October 2026 and currently includes 28 evaluated models.