DeepSeek·released 2025-05-281 source
DeepSeek-R1 (May 2025) benchmark scores: 12 benchmarks tracked. 25% of its scores are independently reproduced — the rest are self-reported or unverified.
Leads on 0 of 12 benchmarks·best result 96.64 on MATH level 5 (reproduced)·3 of 12 independently reproduced
| Benchmark | Score | Evidence status | Measured |
|---|---|---|---|
| MATH level 5 | 96.64accuracy (%) | reproduced· optimizedT1 | 2025-05-28Epoch AI |
| Lech Mazur Writing | 81.9accuracy (%) | unverifiedT1 | 2025-05-28lechmazur/writing Github repository |
| Fiction.LiveBench | 75accuracy (%) | unverifiedT1 | 2025-05-28Fiction.live leaderboard |
| Aider polyglot | 71.4accuracy (%) | unverified· optimizedT1 | 2025-05-28https://aider.chat/docs/leaderboards/ |
| GPQA diamond | 68.43accuracy (%) | reproduced· optimizedT1 | 2025-05-28Epoch AI |
| OTIS Mock AIME 2024-2025 | 66.36accuracy (%) | reproduced· optimizedT1 | 2025-05-28Epoch AI |
| METR Time Horizons | 53.78accuracy (%) | unverifiedT1 | 2025-05-28METR - Measuring AI Ability to Complete Long Tasks |
| WeirdML | 41.63accuracy (%) | unverifiedT1 | 2025-05-28https://htihle.github.io/weirdml.html |
| DeepResearch Bench | 35.1accuracy (%) | unverifiedT1 | 2025-05-28DeepResearchBench Leaderboard |
| SimpleBench | 28.96accuracy (%) | unverifiedT1 | 2025-05-28SimpleBench Leaderboard |
| ARC-AGI | 21.21accuracy (%) | unverified· optimizedT1 | 2025-05-28https://arcprize.org/leaderboard |
| ARC-AGI-2 | 1.12accuracy (%) | unverifiedT2 | 2025-05-28 |
Claims drawn from cited facts, not live model generation.
This model comes from DeepSeek and dates to 2025; with no parameter count provided, its scale cannot be characterized without inventing a number.
2 cited facts
This model is scored on 12 tracked benchmarks, yet it currently holds no top score on any of them. Its results have been independently reproduced, so outside evaluation backs the scores up rather than leaving them as vendor-claimed figures.
3 cited facts
On average, this model trails the leader by 35.08 points across its benchmarks — a score gap under the disclosed harness, not a capability claim — placing it in the wide-gap band and well behind the front-of-pack. It holds the top score on 0 benchmarks, indicating no category leadership at present.
2 cited facts
1 facts cross-checked across data sources: 1 single-source.
Source facts, citations, and refresh stamp for this record.
Sources: epoch_benchmarks
DeepSeek-R1 (May 2025) is an AI model developed by DeepSeek, released 2025-05-28. Its benchmark record below tags every score as independently reproduced, self-reported, or unverified.
DeepSeek-R1 (May 2025) has recorded scores on 12 benchmarks, each shown with its evidence status.
DeepSeek-R1 (May 2025) has recorded scores on 12 benchmarks — Lech Mazur Writing, GPQA diamond, MATH level 5, OTIS Mock AIME 2024-2025, WeirdML, SimpleBench, and 6 more. The full table above shows each score with its evidence status.
3 of 12 recorded scores (25%) are independently reproduced rather than self-reported by the lab.
DeepSeek-R1 (May 2025) has 12 tracked claims: 3 independently reproduced, 9 unverified.