AIME 2026

AIME I and II 2026, evaluated on release by MathArena.

Published by

MathArena

Category

Maths

Index weight

1.0

Models

31

Data as of

9 Sept 2026

MIT — eth-sri/matharena.

1. 1GPT-5.5100.0%
2. 2Claude Opus 4.8100.0%
3. 3GPT-5.499.2%
4. 4Gemini 3.1 Pro98.3%
5. 5GPT-5.298.3%
6. 6Claude Opus 4.696.7%
7. 7Gemini 3.6 Flash96.7%
8. 8GLM-596.7%
9. 9Gemini 3 Flash96.7%
10. 10DeepSeek V4 Pro96.7%
11. 11Step 3.5 Flash96.7%
12. 12Kimi K396.7%
13. 13GLM-5.195.8%
14. 14Claude Opus 4.795.8%
15. 15Kimi K2.595.8%

SearchAll providersAlibabaAnthropicDeepSeekGoogleLM-ProversMoonshot AINVIDIAOpenAIStepFunxAIZhipu AIOpen weights31 of 31

#	Model	Score ↓	Run ↕	Index ↕	Published ↕
1	GPT-5.5OpenAI	100.0%	–	63.7	–
2	Claude Opus 4.8Anthropic	100.0%	–	59.4	–
3	GPT-5.4OpenAI	99.2%	–	61.3	–
4	Gemini 3.1 ProGoogle	98.3%	–	63.9	–
5	GPT-5.2OpenAI	98.3%	–	55.5	–
6	Claude Opus 4.6Anthropic	96.7%	–	61.3	–
7	Gemini 3.6 FlashGoogle	96.7%	Gemini 3.6 Flash	60.7	–
8	GLM-5Zhipu AIopen	96.7%	GLM 5	59.0	–
9	Gemini 3 FlashGoogle	96.7%	Gemini 3 Flash	56.6	–
10	DeepSeek V4 ProDeepSeekopen	96.7%	–	56.1	–
11	Step 3.5 FlashStepFunopen	96.7%	Step 3.5 Flash	54.9	–
12	Kimi K3Moonshot AIopen	96.7%	–	–	–
13	GLM-5.1Zhipu AIopen	95.8%	GLM 5.1	60.1	–
14	Claude Opus 4.7Anthropic	95.8%	–	57.2	–
15	Kimi K2.5Moonshot AIopen	95.8%	–	55.3	–
16	DeepSeek V4 FlashDeepSeekopen	95.8%	–	54.7	–
17	Kimi K2.6Moonshot AIopen	95.8%	–	–	–
18	Gemini 3.5 FlashGoogle	95.0%	Gemini 3.5 Flash	60.7	–
19	Step 3.7 FlashStepFunopen	95.0%	Step 3.7 Flash	54.6	–
20	Qwen3.5 397B-A17BAlibabaopen	94.2%	–	58.6	–
21	Grok 4.1xAI	94.2%	–	55.9	–
22	Qwen3.5 35B-A3BAlibabaopen	93.3%	–	53.4	–
23	Qwen3.5-9BAlibabaopen	92.5%	–	47.5	–
24	Gemini 3 ProGoogle	91.7%	–	61.6	–
25	Qwen3.5 27BAlibabaopen	91.7%	–	55.4	–
26	Nemotron 3 SuperNVIDIA	91.7%	–	44.9	–
27	GLM-5.2Zhipu AIopen	90.0%	GLM 5.2	57.4	–
28	Qwen3.5-4BAlibaba	89.7%	–	47.1	–
29	Qwen3 30B A3BAlibabaopen	88.3%	–	45.1	–
30	qwen3-4b-instruct-2507Alibabaopen	82.5%	thinking, 2507	40.1	–
31	LQed NanoLM-Provers	82.5%	–	–	–