American Invitational Mathematics Examination 2024 (AIME 2024)
We show this table for reference; we do not rank on it.
The 2024 edition of AIME, maintaining the same format of 15 challenging mathematics problems with integer answers from 000 to 999.
Benchmark score on AIME 2024 — October 17, 2026
We compile the AIME 2024 rows from provider self-reports and secondary reports. GPT-OSS 120B leads the table at 95.8%, followed by DeepSeek V3.1 (Reasoning) (93.1%) and GPT-OSS 20B (92.1%). We do not use these results to rank models overall.
GPT-OSS 120B
OpenAI
DeepSeek V3.1 (Reasoning)
DeepSeek
GPT-OSS 20B
OpenAI
26 modelsMathematicsRefreshingDisplay onlyUpdated October 17, 2026
Benchmark score table (26 models)
| Rank | Model / configuration | Score | Parameters (B) | Open / closed |
|---|---|---|---|---|
| 1 | GPT-OSS 120BOpenAI | 95.8% | Not reported | Open |
| 2 | DeepSeek V3.1 (Reasoning)DeepSeek | 93.1% | Not reported | Open |
| 3 | GPT-OSS 20BOpenAI | 92.1% | Not reported | Open |
| 4 | Gemini 2.5 ProGoogle | 92% | Not reported | Closed |
| 5 | DeepSeek-R1-0528DeepSeek | 91.4% | Not reported | Open |
| 6 | Ministral 3 14B (Reasoning)Mistral | 89.8% | Not reported | Open |
| 7 | Grok 3 MinixAI | 89.5% | Not reported | Closed |
| 8 | o3-miniOpenAI | 87.3% | Not reported | Closed |
| 9 | MiniMax M1 80kMiniMax | 86% | Not reported | Closed |
| 10 | Qwen3 235B-A22B (Reasoning)Alibaba | 85.7% | Not reported | Open |
| 11 | Qwen3 32B (Reasoning)Alibaba | 81.4% | Not reported | Open |
| 12 | Phi-4-reasoning-plusMicrosoft | 81.3% | Not reported | Open |
| 13 | DeepSeek-R1DeepSeek | 79.8% | Not reported | Open |
| 14 | QwQ-32BAlibaba | 79.5% | Not reported | Open |
| 15 | Phi-4-reasoningMicrosoft | 75.3% | Not reported | Open |
| 16 | Magistral MediumMistral | 73.6% | Not reported | Closed |
| 17 | Magistral SmallMistral | 70.7% | Not reported | Open |
| 18 | Kimi K2Moonshot AI | 69.6% | Not reported | Closed |
| 19 | DeepSeek V3.1DeepSeek | 66.3% | Not reported | Open |
| 20 | Claude 3.7 Sonnet ThinkingAnthropic | 61.3% | Not reported | Closed |
| 21 | Grok 3 [Beta]xAI | 52.2% | Not reported | Closed |
| 22 | Qwen3 235B-A22BAlibaba | 40.1% | Not reported | Open |
| 23 | DeepSeek V3DeepSeek | 39.2% | Not reported | Open |
| 24 | Qwen3 32BAlibaba | 31% | Not reported | Open |
| 25 | Claude 3.7 SonnetAnthropic | 23.3% | Not reported | Closed |
| 26 | Claude 3.5 HaikuAnthropic | 5.3% | Not reported | Closed |
Among the reported AIME 2024 rows, GPT-OSS 120B is first at 95.8%. The third row is 3.7 points behind. The broader top-10 range is 10.1 points, so the table still separates the published systems.
26 models have been evaluated on AIME 2024. The benchmark falls in the Mathematics category. AIME 2024 is currently displayed for reference but excluded from the scoring formula, so it does not directly affect overall rankings.
About AIME 2024
Year
2024
Tasks
15 problems
Format
Integer answers 000-999
Difficulty
High school olympiad level
AIME 2024 continues the tradition of challenging mathematical reasoning problems. These problems test deep understanding of mathematical concepts and creative problem-solving abilities.
Freshness and provenance
Version
AIME 2024 2024
Refresh cadence
Annual
Staleness state
Refreshing
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 2024 measure?
The 2024 edition of AIME, maintaining the same format of 15 challenging mathematics problems with integer answers from 000 to 999.
Which model scores highest on AIME 2024?
GPT-OSS 120B by OpenAI currently leads with a score of 95.8% on AIME 2024.
How many models are evaluated on AIME 2024?
26 AI models have published results on AIME 2024 in the BenchLM catalog.
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 5,600+ readers.
One email each week. Unsubscribe anytime.