Leipzig Benchmark: leaderboard
The Leipzig Benchmark (Benchmarks in Leipzig workshop, Max Planck Institute for Mathematics in the Sciences, 2026) is a set of 100 research-level mathematics questions with unique, unguessable answers written by 49 research mathematicians. The board counts questions solved out of 100 in a single attempt without tools, checked by LLM judges; higher is better. Questions were kept only if at most three of five frontier models solved them at submission time.
Metric: Questions solved (out of 100) in a single attempt per question: 100 research-level mathematics questions with unique, unguessable answers from 49 mathematicians (Benchmarks in Leipzig, MPI MiS, 2026), API runs with web search and code tools disabled, errors or timeouts rerun up to three times, answers checked by GPT-5.5 and Gemini 3.1 Pro judges; questions were accepted only if at most three of the five project-active models solved them; higher is better. Source: arxiv.org. Saturation forecast: Around March 2027. 10 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-5.5 (xHigh) | 44 | #49 (GPT-5.5) |
| 2 | GPT-5.4 | 28 | #123 |
| 3 | Gemini 3.1 Pro (Preview) (Medium) | 15 | #94 (Gemini 3.1 Pro (Preview)) |
| 4 | Claude Opus 4.6 | 14 | #102 |
| 5 | Gemini 3 Pro | 14 | #137 |
| 6 | Claude Opus 4.7 | 13 | #79 |
| 7 | DeepSeek V4 Pro (High) | 10 | #188 (DeepSeek V4 Pro) |
| 8 | DeepSeek V3.2 | 8 | #406 |
| 9 | Grok 4.3 (High) | 6 | #333 (Grok 4.3) |
| 10 | Grok 4.20 | 5 | #297 |
No result here: #1 Claude Fable 5.2, #2 Gemini 4 Carbon, #5 Claude Opus 5.5.
Interactive version: theaggregate.ai/benchmark?slug=leipzig-benchmark · How It Works · Data refreshed daily, snapshot 2026-11-13.