Updated daily at 05:00 UTC
- Released
- 29 Sep 2025
- Status
- Legacy
- Context
- 131.1K
- Max output
- —
- Input per 1M tokens
- —
- Output per 1M tokens
- —
- Open weights
- Yes
- License
- mit
- Parameters
- 671B, 37B active
- Knowledge cutoff
- —
- Reasoning
- Yes
- Modalities
- text → text
Category scores
Out of 100, from independent and third-party results only, by method category-2026-09-29.1.
| Category | Score | Rank | Coverage |
|---|---|---|---|
| Knowledge & reasoning | 34.2 | Not ranked | 1 of 4 |
| Coding & software engineering | 74.2 | Not ranked | 1 of 5 |
| Professional knowledge work | 42.9 | 1 | 1 of 1 |
| Human preference | 43.0 | Not ranked | 2 of 5 |
Benchmark results
Every result we hold for this model, with who ran it and where it was published.
Knowledge & reasoning
| Benchmark | Metric | Value | Provenance | Setting | Evaluated | Source |
|---|---|---|---|---|---|---|
| DTBench (Decision Theory Benchmark) | Accuracy (capability questions) | 87.7% | Third party | thinking | — | https://conceptualreasoning.ai/dtbench via Epoch AI |
| LMCA (Language Model Conceptual Argumentation) | Correlation with expert ratings (Pearson r) | 0.2911 | Third party | thinking | — | https://conceptualreasoning.ai/lmca via Epoch AI |
Coding & software engineering
| Benchmark | Metric | Value | Provenance | Setting | Evaluated | Source |
|---|---|---|---|---|---|---|
| Aider Polyglot | Percent correct | 74.2% | Third party | thinking | — | Aider LLM Leaderboards via Epoch AI |
| Aider Polyglot | Percent correct | 70.2% | Third party | — | — | Aider LLM Leaderboards via Epoch AI |
| WeirdML (v2) | Average accuracy | 39.5% | Third party | thinking | — | https://htihle.github.io/weirdml.html via Epoch AI |
Professional knowledge work
| Benchmark | Metric | Value | Provenance | Setting | Evaluated | Source |
|---|---|---|---|---|---|---|
| TheAgentCompany | Tasks resolved | 42.9% | Third party | — | — | TheAgentCompany leaderboard via Epoch AI |
Long context
| Benchmark | Metric | Value | Provenance | Setting | Evaluated | Source |
|---|---|---|---|---|---|---|
| CL-bench | Tasks solved | 13.2% | Source not stated | thinking | — | Epoch AI via Epoch AI |
| CL-bench Life | Tasks solved | 9.5% | Source not stated | thinking | — | Epoch AI via Epoch AI |
| Fiction.LiveBench | % correct at 16K tokens (Epoch's column) | 83.3% | Third party | high | — | Fiction.live leaderboard via Epoch AI |
| Fiction.LiveBench | % correct at 16K tokens (Epoch's column) | 52.8% | Third party | — | — | Fiction.live leaderboard via Epoch AI |
Human preference
| Benchmark | Metric | Value | Provenance | Setting | Evaluated | Source |
|---|---|---|---|---|---|---|
| Code Arena: WebDev (arena.ai) | Arena score (Elo scale) | 1272 | Independent | — | 25 Sep 2026 | Arena |
| Code Arena: WebDev (arena.ai) | Votes | 3972 | Independent | — | 25 Sep 2026 | Arena |
| Text Arena (arena.ai) | Arena score (Elo scale) | 1425 | Independent | thinking | 25 Sep 2026 | Arena |
| Text Arena (arena.ai) | Arena score (Elo scale) | 1422 | Independent | — | 25 Sep 2026 | Arena |
| Text Arena (arena.ai) | Votes | 11947 | Independent | — | 25 Sep 2026 | Arena |
| Text Arena (arena.ai) | Votes | 8942 | Independent | thinking | 25 Sep 2026 | Arena |
Other DeepSeek-V versions
- DeepSeek-V4.1-Flash10 Sep 2026
- DeepSeek-V4-Flash-Vision-Exp21 Aug 2026
- DeepSeek-V4-Flash24 Apr 2026
- DeepSeek-V4-Pro24 Apr 2026
- DeepSeek-V3.21 Dec 2025
- DeepSeek-V3.2-Speciale1 Dec 2025
- DeepSeek-V3.1-Terminus22 Sep 2025
- DeepSeek-V3.121 Aug 2025
- DeepSeek-V326 Dec 2024
- DeepSeek-V2.55 Sep 2024
- DeepSeek-V2-Lite16 May 2024
- DeepSeek-V26 May 2024