AIME 2026 benchmark

14 AI models ranked on AIME 2026. Each row shows the model’s score next to its cheapest API price per 1M tokens where it has one, so you can weigh quality against cost.

Top model today

Step 3.5 Flash 2603 (Stepfun) leads at 97%

Data via MathArena →

Models ranked

on this benchmark

Top score

97%

current leader

Type

Single eval

benchmark

Updated

15 July 2026

last source fetch

Cost vs AIME 2026

9 priced models

Most attractive quadrant (cheap + high)

= reasoning model (extended thinking)

Each labelled name links to that model’s page. Curated to the cost-efficiency frontier, top scorers and cheapest so labels stay readable; the full field is in the leaderboard table below (and via “select all”). Reasoning models (extended “thinking” before answering) carry a ring around the dot.

Models ranked on AIME 2026

highest score first · 14 models

Maker

#
1	Step 3.5 Flash 2603	Stepfun	97%	n/a	n/a
2	Kimi K2.6	Moonshot	96%	n/a	n/a
3	Kimi K2.5	Moonshot	96%	$0.50	$2.80
4	GLM 5	Zhipu	96%	$0.80	$2.56
5	Ring-2.6-1T	Inclusionai	96%	n/a	n/a
6	GLM 5p1	Zhipu	95%	$1.40	$4.40
7	Qwen3.6 27B	Alibaba	94%	$0.15	$0.50
8	Qwen3.5 397B A17B	Alibaba	93%	$0.60	$3.60
9	Qwen3.6 35B A3B	Alibaba	93%	$0.14	$0.45
10	EXAONE 4.5 33B	Lg-ai-research	93%	n/a	n/a
11	Qwen3.5-27B	Alibaba	91%	$0.30	$2.40
12	Nemotron 3 120B A12b	NVIDIA	90%	$0.50	$1.50
13	Qwen3 30B A3B Thinking 2507	Alibaba	88%	$0.90	$0.90
14	Ling 2.6 Flash	Inclusionai	74%	n/a	n/a

Prices are on-demand list rates. Enterprise commitments can be materially lower. Cost per task uses best-effort token usage.

Frequently asked questions

What is the AIME 2026 benchmark?

The 2026 American Invitational Mathematics Examination (AIME I and II), used as a maths benchmark on problems released after most models' training cutoffs. Each contest is 15 integer-answer problems, scored on accuracy.

Which AI model scores highest on AIME 2026?

Step 3.5 Flash 2603 (Stepfun) leads with 97%.

How many models are ranked on AIME 2026?

14 models carry a AIME 2026 score. Scores come from Hugging Face leaderboards; where a model has a live API price it is verified against LiteLLM and OpenRouter, and models with no current price are listed without one.

Other benchmarks

compare the same models on a different eval

Benchmark scores via Hugging Face leaderboards · pricing verified against LiteLLM + OpenRouter · refreshed daily · updated 15 July 2026

Every weekday

AI moves fast. Here's your debrief.

News, analysis, tools, and more.

For people who build with AI