Vals AIME (AIME)
We show this table for reference; we do not rank on it.
Challenging national math exam given to top high-school students
Evaluation results published by Vals AI. Published leaderboard aggregates and evaluation settings only. Dataset contents are not included.
AIME score on AIME — April 16, 2026
We mirror the published aime score view for AIME. Gemini 3.1 Pro Preview leads the public snapshot at 98.13%, followed by GPT-5.2 (96.88%) and Muse Spark (96.88%). We do not use these results to rank models overall.
Gemini 3.1 Pro Preview
high reasoning
GPT-5.2
OpenAI
xhigh reasoning
Muse Spark
Meta
96 modelsMathematicsCurrentDisplay onlyUpdated April 16, 2026
AIME score table (96 models)
| Rank | Model / configuration | Score | Parameters (B) | Open / closed |
|---|---|---|---|---|
| 1 | Gemini 3.1 Pro PreviewGooglehigh reasoning | 98.13% | Not reported | Unknown |
| 2 | GPT-5.2OpenAIxhigh reasoning | 96.88% | Not reported | Closed |
| 3 | Muse SparkMeta | 96.88% | Not reported | Closed |
| 4 | Gemini 3 Pro PreviewGooglehigh reasoning | 96.68% | Not reported | Unknown |
| 5 | GPT-5.4OpenAIxhigh reasoning | 96.67% | Not reported | Closed |
| 6 | Grok 4.20 0309 ReasoningxAI | 96.46% | Not reported | Unknown |
| 7 | Claude Opus 4.7Anthropic | 96.25% | Not reported | Closed |
| 8 | Gemini 3 Flash PreviewGooglehigh reasoning | 95.63% | Not reported | Unknown |
| 9 | Claude Opus 4.6 (Adaptive)Anthropic | 95.63% | Not reported | Closed |
| 10 | GPT-5.4 miniOpenAIxhigh reasoning | 95.63% | Not reported | Closed |
| 11 | Kimi K2.5 ThinkingMoonshot AI | 95.63% | Not reported | Unknown |
| 12 | Claude Opus 4.5 ThinkingAnthropic | 95.42% | Not reported | Closed |
| 13 | Qwen3.6 PlusAlibaba | 94.58% | Not reported | Closed |
| 14 | GPT-5OpenAIhigh reasoning | 93.37% | Not reported | Unknown |
| 15 | GPT-5.1OpenAIhigh reasoning | 93.33% | Not reported | Closed |
| 16 | GLM-4.7Z.AI | 93.33% | Not reported | Open |
| 17 | GLM-4.6Z.AI | 92.71% | Not reported | Open |
| 18 | GPT-OSS 120BOpenAI | 92.60% | Not reported | Open |
| 19 | Qwen3.5 FlashAlibaba | 92.50% | Not reported | Closed |
| 20 | Claude Sonnet 4.6Anthropic | 92.29% | Not reported | Closed |
| 21 | Grok 4.1 Fast (Reasoning)xAI | 91.88% | Not reported | Closed |
| 22 | GLM-5.1Z.AI | 91.88% | Not reported | Open |
| 23 | GLM 5 ThinkingZhipu AI | 91.67% | Not reported | Unknown |
| 24 | GPT-5 miniOpenAIhigh reasoning | 91.46% | Not reported | Closed |
| 25 | Grok 4 Fast (Reasoning)xAI | 91.25% | Not reported | Closed |
| 26 | MiniMax M2.7MiniMax | 91.04% | Not reported | Open |
| 27 | Grok 4 0709xAI | 90.56% | Not reported | Unknown |
| 28 | GPT-5.4 nanoOpenAIhigh reasoning | 88.75% | Not reported | Closed |
| 29 | MiniMax M2.5MiniMax | 88.75% | Not reported | Closed |
| 30 | Claude Sonnet 4.5 ThinkingAnthropic | 88.19% | Not reported | Closed |
| 31 | GLM-4.5Z.AI | 86.67% | Not reported | Closed |
| 32 | o3-miniOpenAIhigh reasoning | 86.46% | Not reported | Closed |
| 33 | Qwen3.5 Plus ThinkingAlibaba | 86.04% | Not reported | Unknown |
| 34 | GPT-OSS 20BOpenAI | 86.04% | Not reported | Open |
| 35 | Gemini 2.5 Pro Exp 03 25Google | 85.83% | Not reported | Unknown |
| 36 | Kimi K2 ThinkingMoonshot AI | 85.42% | Not reported | Unknown |
| 37 | o3OpenAIhigh reasoning | 85.28% | Not reported | Closed |
| 38 | Grok 3 Mini Fast High ReasoningxAIhigh reasoning | 85.00% | Not reported | Unknown |
| 39 | DeepSeek V3p2 ThinkingFireworks AIhigh reasoning | 84.58% | Not reported | Unknown |
| 40 | Qwen3 235B-A22BAlibaba | 83.96% | Not reported | Open |
| 41 | o4-miniOpenAIhigh reasoning | 83.67% | Not reported | Closed |
| 42 | Magistral Medium 2509Mistral AI | 83.54% | Not reported | Unknown |
| 43 | Gemini 3.1 Flash Lite PreviewGooglehigh reasoning | 83.33% | Not reported | Unknown |
| 44 | Claude Haiku 4.5 ThinkingAnthropic | 82.71% | Not reported | Closed |
| 45 | GPT-5 nanoOpenAIhigh reasoning | 81.18% | Not reported | Closed |
| 46 | Qwen3 MaxAlibaba | 81.04% | Not reported | Closed |
| 47 | Magistral Small 2509Mistral AI | 80.68% | Not reported | Unknown |
| 48 | Claude Opus 4.1 20250805 ThinkingAnthropic | 78.18% | Not reported | Unknown |
| 49 | MiniMax M2.1MiniMax | 77.92% | Not reported | Unknown |
| 50 | Claude Opus 4.5Anthropic | 76.88% | Not reported | Closed |
| 51 | Claude Sonnet 4 20250514 ThinkingAnthropic | 76.25% | Not reported | Unknown |
| 52 | DeepSeek-R1DeepSeek | 73.96% | Not reported | Open |
| 53 | o1OpenAIhigh reasoning | 71.46% | Not reported | Closed |
| 54 | Grok 3 Mini Fast Low ReasoningxAIlow reasoning | 70.63% | Not reported | Unknown |
| 55 | DeepSeek V3p2Fireworks AInone reasoning | 64.79% | Not reported | Unknown |
| 56 | Moonshotai Kimi K2 InstructTogether AI | 62.71% | Not reported | Unknown |
| 57 | Qwen3 Max PreviewAlibaba | 60.69% | Not reported | Unknown |
| 58 | Grok 3xAI | 58.75% | Not reported | Unknown |
| 59 | Langston Nim Nvidia Llama 3.3 Nemotron Super 49b V1 42e84561 ThinkingTogether AI | 53.54% | Not reported | Unknown |
| 60 | DeepSeek V3 0324DeepSeek | 52.20% | Not reported | Open |
| 61 | Gemini 2.5 Flash Preview 09 2025 ThinkingGoogle | 51.46% | Not reported | Unknown |
| 62 | Gemini 2.5 Flash Preview 09 2025Google | 49.79% | Not reported | Unknown |
| 63 | GPT-4.1 miniOpenAIhigh reasoning | 49.38% | Not reported | Closed |
| 64 | Claude 3.7 Sonnet ThinkingAnthropic | 44.58% | Not reported | Closed |
| 65 | Claude Opus 4.1Anthropic | 44.24% | Not reported | Unknown |
| 66 | Mistral Large 2512Mistral AI | 42.92% | Not reported | Unknown |
| 67 | Mistral Medium 2505Mistral AI | 42.29% | Not reported | Unknown |
| 68 | Gemini 2.5 Flash Lite Preview 09 2025 ThinkingGoogle | 42.08% | Not reported | Unknown |
| 69 | Claude Opus 4Anthropic | 41.25% | Not reported | Unknown |
| 70 | GPT-4.1OpenAIhigh reasoning | 39.58% | Not reported | Closed |
| 71 | Claude Sonnet 4Anthropic | 38.54% | Not reported | Unknown |
| 72 | Grok 4 Fast Non ReasoningxAI | 33.33% | Not reported | Unknown |
| 73 | Gemini 2.0 Flash 001Google | 29.79% | Not reported | Unknown |
| 74 | DeepSeek V3DeepSeek | 27.50% | Not reported | Open |
| 75 | Grok 4.1 Fast Non ReasoningxAI | 27.08% | Not reported | Unknown |
| 76 | GPT-4.1 nanoOpenAIhigh reasoning | 26.46% | Not reported | Closed |
| 77 | Gemini 2.5 Flash Lite Preview 09 2025Google | 26.25% | Not reported | Unknown |
| 78 | Llama4 Maverick Instruct BasicFireworks AI | 25.21% | Not reported | Unknown |
| 79 | Claude 3.7 SonnetAnthropic | 22.29% | Not reported | Closed |
| 80 | Meta Llama Llama 4 Scout 17B 16E InstructTogether AI | 18.96% | Not reported | Unknown |
| 81 | Gemini 1.5 Pro 002Google | 18.75% | Not reported | Unknown |
| 82 | Gemini 1.5 Flash 002Google | 17.29% | Not reported | Unknown |
| 83 | Meta Llama Llama 3.3 70B Instruct TurboTogether AI | 16.04% | Not reported | Unknown |
| 84 | Grok 2 1212xAI | 15.21% | Not reported | Unknown |
| 85 | GPT-4oOpenAIhigh reasoning | 13.96% | Not reported | Closed |
| 86 | Command A 03 2025Cohere | 13.33% | Not reported | Unknown |
| 87 | GPT-4oOpenAIhigh reasoning | 11.88% | Not reported | Closed |
| 88 | GPT-4o miniOpenAIhigh reasoning | 11.46% | Not reported | Closed |
| 89 | Claude 3.5 SonnetAnthropic | 10.00% | Not reported | Closed |
| 90 | Langston Nim Nvidia Llama 3.3 Nemotron Super 49b V1 42e84561Together AI | 9.38% | Not reported | Unknown |
| 91 | Mistral Large 2411Mistral AI | 9.17% | Not reported | Unknown |
| 92 | Mistral Small 2402Mistral AI | 5.63% | Not reported | Unknown |
| 93 | Mistral Small 2503Mistral AI | 3.54% | Not reported | Unknown |
| 94 | Claude 3.5 HaikuAnthropic | 3.33% | Not reported | Closed |
| 95 | Jamba Mini 1.6AI21 Labs | 0.42% | Not reported | Unknown |
| 96 | Jamba Large 1.6AI21 Labs | 0.42% | Not reported | Unknown |
How AIME is shown here
BenchLM mirrors the public Vals AI AIME leaderboard captured from https://www.vals.ai/benchmarks/aime and updated by Vals on April 16, 2026. The snapshot preserves overall scores, uncertainty, latency, cost-per-test metadata, and task-level scores where Vals publishes them.
AIME is display only on BenchLM. Vals proprietary or Vals-hosted aggregate views are useful context, but BenchLM does not use them as weighted ranking inputs or as a replacement for benchmark-native source records.
Snapshot
The published AIME snapshot places Gemini 3.1 Pro Preview first at 98.13%. The third row is 1.25 points behind. The broader top-10 range is 2.50 points, so many of the published results sit in a relatively narrow band.
96 models have been evaluated on AIME. The benchmark falls in the Mathematics category. We keep external benchmark mirrors separate from the weighted global scoring system, so these results remain source-specific evidence. AIME is currently displayed for reference but excluded from the scoring formula, so it does not directly affect overall rankings.
About AIME
Year
2026
Tasks
AIME math problems
Format
Accuracy score
Difficulty
Competition math
BenchLM mirrors the public Vals AI AIME leaderboard as display-only external evidence. The captured snapshot preserves overall scores, task-level scores where Vals publishes them, uncertainty, latency, and cost-per-test metadata. It is excluded from BenchLM weighted rankings.
Freshness and provenance
Version
AIME 2026
Refresh cadence
Quarterly
Staleness state
Current
Question availability
Public benchmark set
BenchLM uses freshness metadata to decide whether a benchmark should still be treated as a strong differentiator, a benchmark to watch, or a display-only reference. For the full scoring policy, see the BenchLM methodology page.
Questions
What does AIME measure?
Challenging national math exam given to top high-school students
Which model leads the published AIME snapshot?
Gemini 3.1 Pro Preview currently leads the published AIME snapshot with 98.13% aime score. BenchLM shows this benchmark for display only and does not use it in overall rankings.
How many models are evaluated on AIME?
The April 16, 2026 snapshot contains 96 AI models.
Know when it’s worth switching models
The model to choose, the cheaper alternative, and the release we would wait on.
Read a sample issueJoin 7,000+ subscribers.
One email each week. Unsubscribe anytime.