Benchmark · Reasoning and knowledge

OTIS Mock AIME 2024-2025

45 competition maths problems from three OTIS mock AIME exams, each with an integer answer from 0 to 999; harder than MATH Level 5 and easier than FrontierMath.

Score
% of problems solved, higher is better
Run by
Epoch AI
Status
Current, newest result 19 November 2026
Updated
as models are released
Ceiling
The best result is within five points of a perfect score, so the test no longer tells the top models apart
Data
the published file, CC BY 4.0, read 20 November 2026
Method
epoch.ai/benchmarks/otis-mock-aime-2024-2025

Results

49 models we list, each at its best setting, best first.

#ModelScoreRangeWho ran itSettingDate
1Gemini 4 Neon B100.0%Independenteffort xhigh2026-11-19source
1Gemini 4 Hydrogen B100.0%Independenteffort xhigh2026-11-13source
1Kimi K3.1 Pro D100.0%Independenteffort max2026-11-06source
1Gemini 4 Carbon B100.0%Independenteffort high2026-11-02source
1GPT-6.2 Astra AA100.0%Independenteffort max2026-10-27source
1GPT-6.1 Terra AA100.0%Independenteffort max2026-10-26source
1Gemini 4 Argon B100.0%Independenteffort high2026-10-13source
1Claude Sonnet 5.5 A100.0%Independenteffort max2026-09-29source
1GPT-6.1 Sol AA100.0%Independenteffort max2026-09-29source
1Claude Opus 5.5 A100.0%Independenteffort max2026-09-22source
1GPT-6 Sol A100.0%Independenteffort max2026-09-22source
1Claude Fable 5.1 A100.0%Independenteffort max2026-09-01source
1GPT-6 Astra A100.0%Independenteffort max2026-08-30source
14Claude Fable 5.2 A99.7%Independenteffort max2026-11-06source
15Mistral Large 4 C99.4%Independenteffort high2026-11-02source
15DeepSeek V4.1 Flash C99.4%Independenteffort max2026-10-19source
15Muse Wonder D99.4%Independenteffort max2026-10-12source
18Muse Spark 1.3 D99.2%Independenteffort xhigh2026-09-17source
19Claude Haiku 5.5 A98.9%Independenteffort xhigh2026-10-07source
19GPT-6 Luna A98.9%Independenteffort max2026-09-22source
19Gemini 3.8 Flash AA98.9%Independenteffort high2026-09-02source
22Muse Spark Open C98.1%Independenteffort xhigh2026-10-15source
22Grok 4.7 BB98.1%Independenteffort xhigh2026-09-22source
24Kimi K3 B97.2%Independenteffort max2026-07-16source
25DeepSeek V4 Pro D96.7%Independenteffort max2026-06-17source
26Gemini 3.1 Pro (preview) AA95.6%Independent2026-02-20source
27DeepSeek-V4-Flash-0731 C94.4%Independenteffort max2026-08-02source
28GLM-5.3-Flash B93.9%Independenteffort max2026-08-26source
29Grok 4.3 BB93.3%Independenteffort high2026-06-17source
30GLM-5.3 B91.1%Independenteffort max2026-08-24source
30Qwen3.6-27B B91.1%Independent2026-08-07source
32Inkling-Small B90.0%Independenteffort xhigh2026-08-14source
33Qwen3.5-397B-A17B B88.9%Independent2026-08-07source
33Inkling B88.9%Independenteffort xhigh2026-08-05source
33gpt-oss-120b BB88.9%Independenteffort high2025-12-11source
36Nemotron 3 Ultra 550B A55B B86.7%Independent2026-08-10source
36Qwen3.6-35B-A3B B86.7%Independent2026-08-07source
38Gemma 4 26B A4B IT BB82.2%Independenteffort minimal2026-08-28source
39Gemini 3.1 Flash-Lite B80.0%Independenteffort high2026-08-06source
40Gemma 4 31B IT BB73.3%Independenteffort minimal2026-08-06source
41MiniMax M3 C71.1%Independent2026-08-10source
41Gemini 3.5 Flash-Lite AA71.1%Independenteffort high2026-08-06source
43gpt-oss-20b B65.3%Independenteffort medium2026-08-27source
44Qwen3-30B-A3B-Instruct-2507 C62.2%Independent2026-08-30source
45Qwen3.5-9B C61.7%Independenteffort none2026-08-27source
46GLM-4.7-Flash B58.3%Independent2026-08-30source
47Qwen3.5-4B C55.8%Independenteffort none2026-08-28source
48Qwen3-4B-Instruct-2507 C52.2%Independent2026-08-28source
49Llama 3.3 70B B5.1%Independent2025-02-25source
Every setting, 68 rows
ModelAs the leaderboard names itScoreWho ran itSettingDate
Gemini 4 Neongemini-4-neon100.0%Independenteffort xhigh2026-11-19source
Gemini 4 Hydrogengemini-4-hydrogen100.0%Independenteffort xhigh2026-11-13source
Kimi K3.1 Prokimi-k3.1-pro-together100.0%Independenteffort max2026-11-06source
Gemini 4 Carbongemini-4-carbon100.0%Independenteffort high2026-11-02source
GPT-6.2 Astragpt-6.2-astra100.0%Independenteffort max2026-10-27source
GPT-6.1 Terragpt-6.1-terra100.0%Independenteffort max2026-10-26source
Gemini 4 Argongemini-4-argon100.0%Independenteffort high2026-10-13source
Claude Sonnet 5.5claude-sonnet-5-5100.0%Independenteffort max2026-09-29source
GPT-6.1 Solgpt-6.1-sol100.0%Independenteffort max2026-09-29source
Claude Opus 5.5claude-opus-5-5100.0%Independenteffort max2026-09-22source
GPT-6 Solgpt-6-sol100.0%Independenteffort max2026-09-22source
Claude Fable 5.1claude-fable-5-1100.0%Independenteffort max2026-09-01source
GPT-6 Astragpt-6-astra100.0%Independenteffort max2026-08-30source
Claude Fable 5.2claude-fable-5-299.7%Independenteffort max2026-11-06source
Mistral Large 4mistral-large-4-0-oss99.4%Independenteffort high2026-11-02source
DeepSeek V4.1 Flashdeepseek-v4.1-flash-101799.4%Independenteffort max2026-10-19source
Muse Wondermuse-wonder-1.099.4%Independenteffort max2026-10-12source
Muse Spark 1.3muse-spark-1.399.2%Independenteffort xhigh2026-09-17source
Claude Haiku 5.5claude-haiku-5-598.9%Independenteffort xhigh2026-10-07source
GPT-6 Lunagpt-6-luna98.9%Independenteffort max2026-09-22source
Gemini 3.8 Flashgemini-3.8-flash98.9%Independenteffort high2026-09-02source
DeepSeek V4.1 Flashdeepseek-v4.1-flash98.3%Independenteffort max2026-10-08source
Muse Spark OpenMuse-Spark-Open-816B-A42B98.1%Independenteffort xhigh2026-10-15source
Grok 4.7grok-4.798.1%Independenteffort xhigh2026-09-22source
Claude Haiku 5.5claude-haiku-5-597.2%Independenteffort high2026-10-07source
Kimi K3kimi-k397.2%Independenteffort max2026-07-16source
DeepSeek V4 Prodeepseek-v4-pro96.7%Independenteffort max2026-06-17source
Gemini 3.1 Pro (preview)gemini-3.1-pro-preview95.6%Independent2026-02-20source
DeepSeek V4 Prodeepseek-v4-pro95.6%Independenteffort high2026-08-06source
Gemini 3.1 Pro (preview)gemini-3.1-pro-preview95.6%Independenteffort high2026-08-06source
DeepSeek-V4-Flash-0731deepseek-v4-flash-073194.4%Independenteffort max2026-08-02source
GLM-5.3-Flashglm-5.3-flash93.9%Independenteffort max2026-08-26source
Kimi K3kimi-k393.3%Independenteffort high2026-08-07source
Grok 4.3grok-4.393.3%Independenteffort high2026-06-17source
GLM-5.3glm-5.391.1%Independenteffort max2026-08-24source
Qwen3.6-27Bqwen3.6-27b91.1%Independent2026-08-07source
Inkling-Smallinkling-small90.0%Independenteffort xhigh2026-08-14source
Qwen3.5-397B-A17Bqwen3.5-397b-a17b88.9%Independent2026-08-07source
InklingInkling88.9%Independenteffort xhigh2026-08-05source
gpt-oss-120bopenai/gpt-oss-120b88.9%Independenteffort high2025-12-11source
Nemotron 3 Ultra 550B A55Bnemotron-3-ultra86.7%Independent2026-08-10source
Qwen3.6-35B-A3Bqwen3.6-35b-a3b86.7%Independent2026-08-07source
Gemma 4 26B A4B ITgemma-4-26b-a4b-it82.2%Independenteffort minimal2026-08-28source
Qwen3.5-397B-A17Bqwen3.5-397b-a17b82.2%Independenteffort none2026-08-07source
Gemini 3.1 Flash-Litegemini-3.1-flash-lite80.0%Independenteffort high2026-08-06source
Gemma 4 31B ITgemma-4-31b-it73.3%Independenteffort minimal2026-08-06source
MiniMax M3MiniMax-M371.1%Independent2026-08-10source
Gemini 3.5 Flash-Litegemini-3.5-flash-lite71.1%Independenteffort high2026-08-06source
Kimi K3kimi-k368.9%Independenteffort low2026-08-07source
Qwen3.6-35B-A3Bqwen3.6-35b-a3b68.9%Independenteffort none2026-08-07source
Qwen3.6-27Bqwen3.6-27b66.7%Independenteffort none2026-08-07source
gpt-oss-20bgpt-oss-20b65.3%Independenteffort medium2026-08-27source
Qwen3-30B-A3B-Instruct-2507qwen3-30b-a3b-instruct-250762.2%Independent2026-08-30source
Qwen3.5-9Bqwen3.5-9B61.7%Independenteffort none2026-08-27source
Gemini 3.5 Flash-Litegemini-3.5-flash-lite60.0%Independenteffort low2026-08-06source
GLM-4.7-Flashglm-4.7-flash58.3%Independent2026-08-30source
Qwen3.5-4Bqwen3.5-4b55.8%Independenteffort none2026-08-28source
Qwen3-4B-Instruct-2507qwen3-4b-instruct-250752.2%Independent2026-08-28source
Gemini 3.5 Flash-Litegemini-3.5-flash-lite51.1%Independenteffort minimal2026-08-06source
gpt-oss-20bgpt-oss-20b50.8%Independenteffort high2026-08-27source
DeepSeek V4 Prodeepseek-v4-pro46.7%Independenteffort none2026-08-06source
Gemini 3.1 Flash-Litegemini-3.1-flash-lite44.4%Independenteffort low2026-08-06source
Qwen3.5-9Bqwen3.5-9B42.8%Independent2026-08-27source
gpt-oss-20bgpt-oss-20b40.3%Independenteffort low2026-08-27source
Gemini 3.1 Flash-Litegemini-3.1-flash-lite37.8%Independenteffort minimal2026-08-06source
MiniMax M3MiniMax-M326.7%Independenteffort none2026-08-10source
GLM-4.7-Flashglm-4.7-flash25.0%Independenteffort none2026-08-28source
Llama 3.3 70BLlama-3.3-70B-Instruct5.1%Independent2025-02-25source

A rank is among the models we list with a result, not a place on the whole leaderboard, and a model offered by several hosts is shown once, on its own listing. These results don't move our grades. Data: Epoch AI, Capabilities and benchmarking (CC BY 4.0), CC BY 4.0.

For companies

Do agents find, use and choose your tools?

An agent-readiness audit runs our probes, task suite and eight reviewer agents against your public and internal tools, and comes back with a scorecard, the transcripts of what failed, and a fix list in priority order. From $2,500, re-run included. We never take payment to move a rank. We do help companies earn one.