benchmark evidence
AIME 2025
AIME 2025 math competition problems from LLM-Stats.
activeupstream source ->
winner on AIME 2025
direct benchmark result, not a broad vertical composite | source row dated 2000-01-01
scored on 2000-01-01 · stale source data (9646d)
latest mapped results | top 20
| # | Model | Score | Evidence | Tested |
|---|---|---|---|---|
| 1 | MoonshotAI: Kimi K2 Thinking | 100.0 | model-only independent_benchmark | 2000-01-01 |
| 2 | OpenAI: GPT-5.2 | 100.0 | model-only independent_benchmark | 2000-01-01 |
| 3 | Anthropic: Claude Opus 4.6 | 99.8 | model-only independent_benchmark | 2000-01-01 |
| 4 | OpenAI: GPT-5 | 94.6 | model-only independent_benchmark | 2000-01-01 |
| 5 | OpenAI: GPT-5.1 | 94.0 | model-only independent_benchmark | 2000-01-01 |
| 6 | DeepSeek: DeepSeek V3.2 | 93.1 | model-only independent_benchmark | 2000-01-01 |
| 7 | OpenAI: o4 Mini | 92.7 | model-only independent_benchmark | 2000-01-01 |
what this result means
AIME 2025 math competition problems from LLM-Stats.
This benchmark contributes direct public evidence. Read its scope before generalizing the result.
A win here is a win on AIME 2025. Broad task pages require independent corroboration before naming a general winner.
source record
category: math
metric: accuracy
matched models: 7
latest source date: 2000-01-01
direction: higher is better