AI models
OTIS Mock AIME: AI model scores
182 AI models on OTIS Mock AIME, per Epoch AI. It is a saturated benchmark: several models from different labs have already hit the ceiling.
OTIS Mock AIME, on the Epoch AI leaderboard (CC BY 4.0, latest version we checked), is a set of math problems in the style of the US AIME olympiad, in a practice version built so the model cannot lean on answers it may already have seen. The score is the share of correct answers. Here are the 182 models in this section with a score on OTIS Mock AIME, per Epoch AI. 16 models score 100%, so the table shows the scores without a rank.
OTIS Mock AIME
100%
Claude Fable 5
models
182
01
| Model | % | Source |
|---|---|---|
| Claude Fable 5Anthropic · closed | 100% top score | Epoch AI |
| Claude Fable 5.1Anthropic · closed | 100% top score | Epoch AI |
| Claude Opus 5.5Anthropic · closed | 100% top score | Epoch AI |
| Claude Sonnet 5.5Anthropic · closed | 100% top score | Epoch AI |
| Gemini 4 ArgonGoogle · closed | 100% top score | Epoch AI |
| Gemini 4 CarbonGoogle · closed | 100% top score | Epoch AI |
| GPT-5.5OpenAI · closed | 100% top score | Epoch AI |
| GPT-5.5 ProOpenAI · closed | 100% top score | Epoch AI |
| GPT-5.6 SolOpenAI · closed | 100% top score | Epoch AI |
| GPT-6 AstraOpenAI · closed | 100% top score | Epoch AI |
| GPT-6 SolOpenAI · closed | 100% top score | Epoch AI |
| GPT-6.1 SolOpenAI · closed | 100% top score | Epoch AI |
| GPT-6.1 TerraOpenAI · closed | 100% top score | Epoch AI |
| GPT-6.2 AstraOpenAI · closed | 100% top score | Epoch AI |
| Kimi K3.1 ProMoonshot AI · open | 100% top score | Epoch AI |
| Qwen3.8 MaxAlibaba · closed | 100% top score | Epoch AI |
| Claude Fable 5.2Anthropic · closed | 99.7% statistical tie with the top score | Epoch AI |
| GPT-5.6 TerraOpenAI · closed | 99.7% statistical tie with the top score | Epoch AI |
| Mistral Large 4Mistral AI · open | 99.4% statistical tie with the top score | Epoch AI |
| Muse WonderMeta · closed | 99.4% statistical tie with the top score | Epoch AI |
| Doubao Seed Evolving (Nov 2026)ByteDance · closed | 99.2% statistical tie with the top score | Epoch AI |
| Grok 4.6xAI · closed | 99.2% statistical tie with the top score | Epoch AI |
| Muse Spark 1.3Meta · closed | 99.2% statistical tie with the top score | Epoch AI |
| Claude Haiku 5.5Anthropic · closed | 98.9% statistical tie with the top score | Epoch AI |
| Claude Opus 5Anthropic · closed | 98.9% statistical tie with the top score | Epoch AI |
| Gemini 3.8 FlashGoogle · closed | 98.9% statistical tie with the top score | Epoch AI |
| GPT-6 LunaOpenAI · closed | 98.9% statistical tie with the top score | Epoch AI |
| Claude Opus 4.8Anthropic · closed | 98.3% statistical tie with the top score | Epoch AI |
| DeepSeek V4.1 FlashDeepSeek · open | 98.3% statistical tie with the top score | Epoch AI |
| GPT-5.6 LunaOpenAI · closed | 98.3% statistical tie with the top score | Epoch AI |
| Grok 4.7xAI · closed | 98.1% statistical tie with the top score | Epoch AI |
| Muse Spark OpenMeta · open | 98.1% statistical tie with the top score | Epoch AI |
| Claude Opus 4.7Anthropic · closed | 97.8% statistical tie with the top score | Epoch AI |
| GPT-5.4OpenAI · closed | 97.8% statistical tie with the top score | Epoch AI |
| Grok 4.5xAI · closed | 97.8% statistical tie with the top score | Epoch AI |
| Gemini 3.7 FlashGoogle · closed | 97.2% statistical tie with the top score | Epoch AI |
| Kimi K3Moonshot AI · open | 97.2% statistical tie with the top score | Epoch AI |
| DeepSeek V4 ProDeepSeek · open | 96.7% statistical tie with the top score | Epoch AI |
| Doubao Seed 2.1 TurboByteDance · closed | 96.1% statistical tie with the top score | Epoch AI |
| GPT-5.2OpenAI · closed | 96.1% statistical tie with the top score | Epoch AI |
| Kimi K2.6Moonshot AI · open | 96.1% statistical tie with the top score | Epoch AI |
| Doubao Seed 2.1 ProByteDance · closed | 95.6% statistical tie with the top score | Epoch AI |
| Gemini 3 FlashGoogle · closed | 95.6% statistical tie with the top score | Epoch AI |
| Gemini 3.1 ProGoogle · closed | 95.6% statistical tie with the top score | Epoch AI |
| Gemini 3.5 FlashGoogle · closed | 95.6% statistical tie with the top score | Epoch AI |
| Kimi K2.7 CodeMoonshot AI · open | 95.6% statistical tie with the top score | Epoch AI |
| Qwen3.7 MaxAlibaba · closed | 95.6% statistical tie with the top score | Epoch AI |
| Claude Sonnet 5Anthropic · closed | 94.7% statistical tie with the top score | Epoch AI |
| Claude Opus 4.6Anthropic · closed | 94.4% statistical tie with the top score | Epoch AI |
| DeepSeek V4 Flash 0731DeepSeek · open | 94.4% statistical tie with the top score | Epoch AI |
| Gemini 3.6 FlashGoogle · closed | 94.2% statistical tie with the top score | Epoch AI |
| Doubao-Seed-2.0 ProByteDance · closed | 93.9% statistical tie with the top score | Epoch AI |
| GLM-5.3-FlashZhipu AI (Z.ai) · open | 93.9% statistical tie with the top score | Epoch AI |
| GLM 5.1Zhipu AI (Z.ai) · open | 93.3% statistical tie with the top score | Epoch AI |
| Grok 4.3 BetaxAI · closed | 93.3% statistical tie with the top score | Epoch AI |
| Qwen3.6 PlusAlibaba · closed | 93.3% statistical tie with the top score | Epoch AI |
| Qwen3.7 PlusAlibaba · closed | 93.3% statistical tie with the top score | Epoch AI |
| Grok 4.20xAI · closed | 92.2% statistical tie with the top score | Epoch AI |
| Kimi K2.5Moonshot AI · open | 92.2% statistical tie with the top score | Epoch AI |
| Gemini 3 ProGoogle · closed | 91.4% statistical tie with the top score | Epoch AI |
| GPT-5OpenAI · closed | 91.4% statistical tie with the top score | Epoch AI |
| Doubao Seed 2.0 Lite (Apr 2026)ByteDance · closed | 91.1% statistical tie with the top score | Epoch AI |
| GLM-5.3Zhipu AI (Z.ai) · open | 91.1% statistical tie with the top score | Epoch AI |
| Qwen 3.6 Max (Preview)Alibaba · undisclosed | 91.1% statistical tie with the top score | Epoch AI |
| Qwen3.6 27BAlibaba · open | 91.1% statistical tie with the top score | Epoch AI |
| GPT-5.4 MiniOpenAI · closed | 88.9% statistical tie with the top score | Epoch AI |
| gpt-oss-120bOpenAI · open | 88.9% statistical tie with the top score | Epoch AI |
| Muse SparkMeta · closed | 88.9% statistical tie with the top score | Epoch AI |
| Qwen3.5 397B A17BAlibaba · open | 88.9% statistical tie with the top score | Epoch AI |
| GPT-5.1OpenAI · closed | 88.6% statistical tie with the top score | Epoch AI |
| DeepSeek V3.2DeepSeek · open | 87.8% statistical tie with the top score | Epoch AI |
| GPT-5.4 NanoOpenAI · closed | 87.8% statistical tie with the top score | Epoch AI |
| GPT-5 MiniOpenAI · closed | 86.7% statistical tie with the top score | Epoch AI |
| Nemotron 3 UltraNVIDIA · open | 86.7% statistical tie with the top score | Epoch AI |
| Qwen3-235B-A22BAlibaba · open | 86.7% statistical tie with the top score | Epoch AI |
| Qwen3.5 Plus 2026-02-15Alibaba · undisclosed | 86.7% statistical tie with the top score | Epoch AI |
| Qwen3.6 35B A3BAlibaba · open | 86.7% statistical tie with the top score | Epoch AI |
| Qwen3.7 FlashAlibaba · closed | 86.7% statistical tie with the top score | Epoch AI |
| GLM-5.2Zhipu AI (Z.ai) · open | 86.4% statistical tie with the top score | Epoch AI |
| Claude Opus 4.5Anthropic · closed | 86.1% statistical tie with the top score | Epoch AI |
| Claude Sonnet 4.6Anthropic · closed | 85.8% statistical tie with the top score | Epoch AI |
| Gemini 2.5 ProGoogle · closed | 84.7% statistical tie with the top score | Epoch AI |
| o3OpenAI · closed | 84.4% statistical tie with the top score | Epoch AI |
| Qwen3.5-FlashAlibaba · undisclosed | 84.4% statistical tie with the top score | Epoch AI |
| Qwen3.6 FlashAlibaba · undisclosed | 84.4% statistical tie with the top score | Epoch AI |
| Grok 4xAI · closed | 84% statistical tie with the top score | Epoch AI |
| GLM 4.7Zhipu AI (Z.ai) · open | 83.3% statistical tie with the top score | Epoch AI |
| Kimi K2 ThinkingMoonshot AI · open | 83.1% statistical tie with the top score | Epoch AI |
| Gemma 4 26B A4BGoogle · open | 82.2% statistical tie with the top score | Epoch AI |
| o4-miniOpenAI · closed | 81.7% statistical tie with the top score | Epoch AI |
| GPT-5 NanoOpenAI · closed | 81.1% statistical tie with the top score | Epoch AI |
| Gemini 3.1 Flash LiteGoogle · closed | 80% statistical tie with the top score | Epoch AI |
| GLM 5Zhipu AI (Z.ai) · open | 80% statistical tie with the top score | Epoch AI |
| Claude Sonnet 4.5Anthropic · closed | 77.8% statistical tie with the top score | Epoch AI |
| Grok-3 minixAI · closed | 77.8% statistical tie with the top score | Epoch AI |
| o3-miniOpenAI · closed | 76.9% statistical tie with the top score | Epoch AI |
| o1OpenAI · closed | 73.3% statistical tie with the top score | Epoch AI |
| Qwen3-MaxAlibaba · closed | 73.3% statistical tie with the top score | Epoch AI |
| Claude Sonnet 4Anthropic · closed | 71.1% statistical tie with the top score | Epoch AI |
| Gemini 3.5 Flash LiteGoogle · closed | 71.1% statistical tie with the top score | Epoch AI |
| MiniMax M3MiniMax · open | 71.1% statistical tie with the top score | Epoch AI |
| Qwen3.5-35B-A3BAlibaba · open | 70% statistical tie with the top score | Epoch AI |
| Doubao-Seed-1.6ByteDance · closed | 69.4% statistical tie with the top score | Epoch AI |
| Claude Opus 4.1Anthropic · closed | 68.9% | Epoch AI |
| GPT-5.5 InstantOpenAI · closed | 68.1% | Epoch AI |
| Qwen3 32BAlibaba · open | 66.9% | Epoch AI |
| Claude Haiku 4.5Anthropic · closed | 66.7% | Epoch AI |
| DeepSeek R1DeepSeek · open | 66.4% | Epoch AI |
| Qwen3 14BAlibaba · open | 66.4% | Epoch AI |
| gpt-oss-20bOpenAI · open | 65.3% | Epoch AI |
| Claude Opus 4Anthropic · closed | 64.4% | Epoch AI |
| Qwen3 30B A3BAlibaba · open | 62.8% | Epoch AI |
| Qwen3.5-9BAlibaba · open | 61.7% | Epoch AI |
| QwQ-32BAlibaba · open | 59.2% | Epoch AI |
| Claude 3.7 SonnetAnthropic · closed | 57.8% | Epoch AI |
| Gemini 2.0 Flash ThinkingGoogle · closed | 57.8% | Epoch AI |
| Qwen3-8BAlibaba · open | 56.1% | Epoch AI |
| Qwen3.5-4BAlibaba · open | 55.8% | Epoch AI |
| DeepSeek-R1-Distill-Qwen-32BDeepSeek · open | 55.6% | Epoch AI |
| Grok 3xAI · closed | 55.6% | Epoch AI |
| DeepSeek-R1-Distill-Llama-70BDeepSeek · open | 51.4% | Epoch AI |
| DeepSeek-R1-Distill-Qwen-14BDeepSeek · open | 50.6% | Epoch AI |
| o1-miniOpenAI · closed | 46.9% | Epoch AI |
| GPT-4.1 MiniOpenAI · closed | 44.7% | Epoch AI |
| GPT-4.1OpenAI · closed | 38.3% | Epoch AI |
| DeepSeek V3DeepSeek · open | 37.8% | Epoch AI |
| GPT-4.5OpenAI · closed | 37.8% | Epoch AI |
| Mistral Medium 3Mistral AI · closed | 32.2% | Epoch AI |
| Gemini 2.0 FlashGoogle · closed | 31.1% | Epoch AI |
| o1-previewOpenAI · closed | 31.1% | Epoch AI |
| Mistral Small 3.2Mistral AI · open | 30.3% | Epoch AI |
| Magistral Small 1.0Mistral AI · open | 30% | Epoch AI |
| GPT-4.1 NanoOpenAI · closed | 28.9% | Epoch AI |
| Magistral Small 1.2Mistral AI · open | 28.1% | Epoch AI |
| Gemini 1.5 ProGoogle · closed | 23.1% | Epoch AI |
| Gemma 3 27BGoogle · open | 22.5% | Epoch AI |
| DeepSeek-R1-Distill-Qwen-1.5BDeepSeek · open | 21.4% | Epoch AI |
| Llama 4 MaverickMeta · open | 20.6% | Epoch AI |
| Qwen-PlusAlibaba · closed | 17.8% | Epoch AI |
| Gemma 3 12BGoogle · open | 16.7% | Epoch AI |
| Gemini 1.5 FlashGoogle · closed | 16.2% | Epoch AI |
| Qwen2.5-MaxAlibaba · closed | 16.1% | Epoch AI |
| Phi 4Microsoft · open | 13.8% | Epoch AI |
| Grok-2xAI · closed | 11.5% | Epoch AI |
| Llama 3.1-405BMeta · open | 9.7% | Epoch AI |
| Claude 3.5 Sonnet (October 2024)Anthropic · closed | 8.5% | Epoch AI |
| Qwen2.5-72BAlibaba · open | 8.1% | Epoch AI |
| Qwen3-1.7BAlibaba · open | 8.1% | Epoch AI |
| Llama 4 ScoutMeta · open | 7.8% | Epoch AI |
| Mistral Large 2Mistral AI · open | 7.8% | Epoch AI |
| Gemma 3 4BGoogle · open | 7.5% | Epoch AI |
| Qwen2.5-32BAlibaba · open | 7.4% | Epoch AI |
| GPT-4o-miniOpenAI · closed | 6.9% | Epoch AI |
| GPT-4 Turbo (Apr 2024)OpenAI · closed | 6.7% | Epoch AI |
| Mistral Small 3Mistral AI · open | 6.7% | Epoch AI |
| Claude 3.5 SonnetAnthropic · closed | 6.5% | Epoch AI |
| Qwen-TurboAlibaba · closed | 6.1% | Epoch AI |
| Mistral Small 3.1Mistral AI · open | 5.8% | Epoch AI |
| Llama 3.3 70BMeta · open | 5.1% | Epoch AI |
| Claude 3 OpusAnthropic · closed | 4.7% | Epoch AI |
| Gemini 1.5 Flash 8BGoogle · closed | 4.6% | Epoch AI |
| Tulu 3 (Tülu 3) 70BAi2 (Allen Institute) · open | 4.4% | Epoch AI |
| Claude 3.5 HaikuAnthropic · closed | 4.3% | Epoch AI |
| Llama 3-70BMeta · open | 4.3% | Epoch AI |
| Llama 3.1-70BMeta · open | 3.6% | Epoch AI |
| Llama 3.2 90BMeta · open | 2.6% | Epoch AI |
| Claude 2Anthropic · closed | 2.5% | Epoch AI |
| Claude 3 SonnetAnthropic · closed | 2.5% | Epoch AI |
| Qwen2.5-7BAlibaba · open | 2.5% | Epoch AI |
| Claude 2.1Anthropic · closed | 1.9% | Epoch AI |
| Llama 3-8BMeta · open | 1.9% | Epoch AI |
| Mistral LargeMistral AI · closed | 1.9% | Epoch AI |
| Claude 3 HaikuAnthropic · closed | 1.8% | Epoch AI |
| Llama 3.1-8BMeta · open | 1.7% | Epoch AI |
| Gemma 2 27BGoogle · open | 1.4% | Epoch AI |
| Gemini 1.0 ProGoogle · closed | 1.1% | Epoch AI |
| Gemma 3 1BGoogle · open | 1.1% | Epoch AI |
| DeepSeek LLM 67BDeepSeek · open | 0.8% | Epoch AI |
| Gemma 2 9BGoogle · open | 0.6% | Epoch AI |
| Llama 3.2 1BMeta · open | 0.6% | Epoch AI |
| Mistral 7BMistral AI · open | 0.3% | Epoch AI |
| Llama 2-70BMeta · open | 0% | Epoch AI |
FAQ
Which model has the top score on OTIS Mock AIME?
16 models score 100% on OTIS Mock AIME, per Epoch AI. It is a saturated benchmark: several models from different labs have already hit the ceiling, and the table shows the scores without a rank.
What does OTIS Mock AIME measure?
OTIS Mock AIME, on the Epoch AI leaderboard (CC BY 4.0, latest version we checked), is a set of math problems in the style of the US AIME olympiad, in a practice version built so the model cannot lean on answers it may already have seen. The score is the share of correct answers.
Which models are in this ranking?
The 182 models in this section with a published score on OTIS Mock AIME, per Epoch AI. No version inherits another's score.
How to read
- Benchmark scores are percentages, and higher is better.
- Rank is shown as a range, «1st of 182». Within the range, the gap to neighboring models falls within the benchmark's margin. The «top score» badge marks the model with the benchmark's top score, and the «statistical tie with the top score» badge marks models whose gap to it is within the margin.
- On a saturated benchmark, the table shows the scores with the badge and without a rank order.
Sources
- Epoch AI, CC BY 4.0, latest version we checked.
- OpenRouter Benchmarks, CC BY 4.0, latest version we checked.
- OpenRouter, dados públicos da API; exports de benchmarks e rankings em CC BY 4.0, latest version we checked.
Dates come from the sources: OpenRouter listing or Hugging Face repo creation, which may differ from the official launch. Family and version were grouped by rule from the name.
How to cite
If you use this data, cite the page and the date you accessed it.
Adriano Meirinho. "OTIS Mock AIME: AI model scores". meirinho.com.br. https://meirinho.com.br/en/ia/modelos/benchmark/otis-aime/. Accessed November 21, 2026.
Third-party data follows the license of each source. The full table is available as CSV and JSON.