OTIS Mock AIME

Mock AIME competition problems from 2024–2025. Run by Epoch AI.

Published by

Epoch AI Benchmarking Hub

Category

Maths

Index weight

0.5

Models

255

Data as of

9 Sept 2026

CC BY 4.0 — Epoch AI, ‘AI Benchmarking Hub’, epoch.ai/benchmarks. Mirrored boards credit their original publishers.

1. 1GPT-6 Astra100.0%
2. 2Claude Fable 5.1100.0%
3. 3GPT-5.6 Sol100.0%
4. 4Claude Fable 5100.0%
5. 5Qwen3 8100.0%
6. 6GPT-5.5100.0%
7. 7GPT-5.5 Pro100.0%
8. 8Claude Fable 599.7%
9. 9GPT-5.6 Terra99.7%
10. 10Grok 4.699.2%
11. 11Claude Opus 598.9%
12. 12Gemini 3.8 Flash98.9%
13. 13DeepSeek V4 Pro98.6%
14. 14Claude Opus 4.898.3%
15. 15GPT-5.6 Luna98.3%

SearchAll providersAlibabaAllen Institute for AIAnthropicDeepSeekGoogleIBMMetaMicrosoftMiniMaxMistral AIMoonshot AINous ResearchNVIDIAOpenAIThinking MachinesxAIZhipu AIOpen weights255 of 255

#	Model	Score ↓	Run ↕	Index ↕	Published ↕
1	GPT-6 AstraOpenAI	100.0%	–	70.6	–
2	Claude Fable 5.1Anthropic	100.0%	–	70.2	–
3	GPT-5.6 SolOpenAI	100.0%	–	65.5	–
4	Claude Fable 5Anthropic	100.0%	–	63.9	–
5	Qwen3 8Alibaba	100.0%	xhigh, 0902	58.0	–
6	GPT-5.5OpenAI	100.0%	–	–	–
7	GPT-5.5 ProOpenAI	100.0%	–	–	–
8	Claude Fable 5Anthropic	99.7%	–	65.8	–
9	GPT-5.6 TerraOpenAI	99.7%	–	58.0	–
10	Grok 4.6xAI	99.2%	–	62.3	–
11	Claude Opus 5Anthropic	98.9%	–	66.8	–
12	Gemini 3.8 FlashGoogle	98.9%	–	60.1	–
13	DeepSeek V4 ProDeepSeekopen	98.6%	max, 0813	56.1	–
14	Claude Opus 4.8Anthropic	98.3%	–	59.4	–
15	GPT-5.6 LunaOpenAI	98.3%	–	56.5	–
16	Claude Opus 4.7Anthropic	97.8%	–	57.2	–
17	Claude Opus 5Anthropic	97.8%	–	70.0	–
18	GPT-5.4OpenAI	97.8%	–	58.8	–
19	Grok 4.6xAI	97.8%	–	58.7	–
20	Grok 4.5xAI	97.8%	–	54.2	–
21	Claude Fable 5Anthropic	97.8%	–	–	–
22	Claude Opus 4.8Anthropic	97.8%	–	–	–
23	Kimi K3Moonshot AIopen	97.2%	–	62.7	–
24	Gemini 3.7 FlashGoogle	97.2%	–	60.8	–
25	Kimi K2.6Moonshot AIopen	96.1%	kimi-k2.6	60.6	–
26	GPT-5.2OpenAI	96.1%	–	55.5	–
27	GPT-5.2OpenAI	96.1%	–	57.3	–
28	Gemini 3.1 ProGoogle	95.6%	–	63.9	–
29	GPT-5.6 SolOpenAI	95.6%	–	64.6	–
30	Gemini 3.1 ProGoogle	95.6%	–	60.7	–
31	Gemini 3.5 FlashGoogle	95.6%	–	60.4	–
32	Deepseek v4 ProDeepSeek	95.6%	–	60.3	–
33	Qwen3 7Alibaba	95.6%	–	58.9	–
34	Gemini 3 FlashGoogle	95.6%	–	58.5	–
35	Kimi K2.7 CodeMoonshot AIopen	95.6%	–	56.2	–
36	GPT-5.4OpenAI	95.6%	–	56.0	–
37	GPT-5.4OpenAI	95.3%	–	61.3	–
38	Claude Sonnet 5Anthropic	94.7%	–	57.3	–
39	Claude Opus 4.6Anthropic	94.4%	–	63.1	–
40	DeepSeek V4 FlashDeepSeekopen	94.4%	max, 0731	54.7	–
41	Gemini 3.6 FlashGoogle	94.2%	–	58.6	–
42	GPT-5.2OpenAI	93.9%	–	58.5	–
43	GLM-5.3-FlashZhipu AIopen	93.9%	–	55.6	–
44	Claude Opus 5Anthropic	93.3%	–	61.7	–
45	GLM-5.1Zhipu AIopen	93.3%	glm-5.1	60.1	–
46	Qwen3.7 PlusAlibaba	93.3%	–	59.7	–
47	Qwen3.6 PlusAlibabaopen	93.3%	–	59.0	–
48	Grok 4.3xAI	93.3%	–	50.1	–
49	Kimi K3Moonshot AIopen	93.3%	–	–	–
50	Gemini 3 FlashGoogle	92.8%	–	56.6	–
51	Grok 4.20xAI	92.2%	–	54.3	–
52	Kimi K2.5Moonshot AIopen	92.2%	fireworks/kimi-k2p5	56.1	–
53	Gemini 3 ProGoogle	91.4%	–	61.6	–
54	GPT-5OpenAI	91.4%	–	54.5	–
55	GLM-5.3Zhipu AIopen	91.1%	–	59.9	–
56	Qwen3 6Alibaba	91.1%	–	59.5	–
57	Claude Opus 4.6Anthropic	91.1%	–	54.1	–
58	Qwen3.6 27BAlibabaopen	91.1%	–	53.2	–
59	TInkling-SmallThinking Machinesopen	90.0%	–	48.5	–
60	Muse SparkMeta	88.9%	–	65.4	–

Show 100 more