DeepSeek-V4：迈向高效百万级 Token 上下文智能
引言
我们推出 DeepSeek-V4 系列的预览版本，包含两款强大的混合专家（MoE）语言模型——参数量 1.6T（激活 49B）的 DeepSeek-V4-Pro 和参数量 284B（激活 13B）的 DeepSeek-V4-Flash——两者均支持百万 token 的上下文长度。
DeepSeek-V4 系列在架构与优化方面引入了多项关键升级：
- 混合注意力架构：我们设计了一种结合压缩稀疏注意力（CSA）与高度压缩注意力（HCA）的混合注意力机制，大幅提升了长上下文效率。在 1M token 上下文设置下，与 DeepSeek-V3.2 相比，DeepSeek-V4-Pro 仅需 27% 的单 token 推理 FLOPs 和 10% 的 KV 缓存。
- 流形约束超连接（mHC）：我们引入 mHC 来增强传统残差连接，在保持模型表达能力的同时，提升了信号在各层间传播的稳定性。
- Muon 优化器：我们采用 Muon 优化器以实现更快的收敛速度和更高的训练稳定性。
我们对两个模型在超过 32T 多样化且高质量的 token 上进行了预训练，随后进行了全面的后训练流程。后训练采用两阶段范式：先通过 SFT 和基于 GRPO 的强化学习独立培养领域专家，再通过在线策略蒸馏进行统一模型整合，将不同领域的独特能力融合到单一模型中。
DeepSeek-V4-Pro-Max 作为 DeepSeek-V4-Pro 的最大推理努力模式，显著提升了开源模型的知识能力，牢固确立了其作为当前最佳开源模型的地位。它在编程基准测试中取得了顶级性能，并在推理和智能体任务上大幅缩小了与领先闭源模型的差距。与此同时，DeepSeek-V4-Flash-Max 在获得更大思考预算时，推理性能可与 Pro 版本媲美，不过由于其参数量较小，在纯知识任务和最复杂的智能体工作流上自然稍逊一筹。
模型下载
| 模型 | #总参数量 | 激活参数量 | 上下文长度 | 精度 | 下载 |
|---|---|---|---|---|---|
| DeepSeek-V4-Flash-Base | 284B | 13B | 1M | FP8 混合 | HuggingFace | ModelScope |
| DeepSeek-V4-Flash | 284B | 13B | 1M | FP4 + FP8 混合* | HuggingFace | ModelScope |
| DeepSeek-V4-Pro-Base | 1.6T | 49B | 1M | FP8 混合 | HuggingFace | ModelScope |
| DeepSeek-V4-Pro | 1.6T | 49B | 1M | FP4 + FP8 混合* | HuggingFace | ModelScope |
*FP4 + FP8 混合：MoE 专家参数使用 FP4 精度；其他大部分参数使用 FP8。
评测结果
基座模型
| 评测基准（指标） | # 样本数 | DeepSeek-V3.2-Base | DeepSeek-V4-Flash-Base | DeepSeek-V4-Pro-Base |
|---|---|---|---|---|
| 架构 | - | MoE | MoE | MoE |
| 激活参数量 | - | 37B | 13B | 49B |
| 总参数量 | - | 671B | 284B | 1.6T |
| 世界知识 | ||||
| AGIEval（精确匹配） | 0-shot | 80.1 | 82.6 | 83.1 |
| MMLU（精确匹配） | 5-shot | 87.8 | 88.7 | 90.1 |
| MMLU-Redux（精确匹配） | 5-shot | 87.5 | 89.4 | 90.8 |
| MMLU-Pro（精确匹配） | 5-shot | 65.5 | 68.3 | 73.5 |
| MMMLU（精确匹配） | 5-shot | 87.9 | 88.8 | 90.3 |
| C-Eval（精确匹配） | 5-shot | 90.4 | 92.1 | 93.1 |
| CMMLU（精确匹配） | 5-shot | 88.9 | 90.4 | 90.8 |
| MultiLoKo（精确匹配） | 5-shot | 38.7 | 42.2 | 51.1 |
| Simple-QA verified（精确匹配） | 25-shot | 28.3 | 30.1 | 55.2 |
| SuperGPQA（精确匹配） | 5-shot | 45.0 | 46.5 | 53.9 |
| FACTS Parametric（精确匹配） | 25-shot | 27.1 | 33.9 | 62.6 |
| TriviaQA（精确匹配） | 5-shot | 83.3 | 82.8 | 85.6 |
| 语言与推理 | ||||
| BBH（精确匹配） | 3-shot | 87.6 | 86.9 | 87.5 |
| DROP（F1） | 1-shot | 88.2 | 88.6 | 88.7 |
| HellaSwag（精确匹配） | 0-shot | 86.4 | 85.7 | 88.0 |
| WinoGrande（精确匹配） | 0-shot | 78.9 | 79.5 | 81.5 |
| CLUEWSC（精确匹配） | 5-shot | 83.5 | 82.2 | 85.2 |
| 代码与数学 | ||||
| BigCodeBench（Pass@1） | 3-shot | 63.9 | 56.8 | 59.2 |
| HumanEval（Pass@1） | 0-shot | 62.8 | 69.5 | 76.8 |
| GSM8K（精确匹配） | 8-shot | 91.1 | 90.8 | 92.6 |
| MATH（精确匹配） | 4-shot | 60.5 | 57.4 | 64.5 |
| MGSM（精确匹配） | 8-shot | 81.3 | 85.7 | 84.4 |
| CMath（精确匹配） | 3-shot | 92.6 | 93.6 | 90.9 |
| 长上下文 | ||||
| LongBench-V2（精确匹配） | 1-shot | 40.2 | 44.7 | 51.5 |
指令模型
DeepSeek-V4-Pro 和 DeepSeek-V4-Flash 均支持三种推理努力模式：
| 推理模式 | 特点 | 典型使用场景 | 响应格式 |
|---|---|---|---|
| 非思考 | 快速、直觉式响应 | 日常常规任务、低风险决策 | </think> 摘要 |
| 高思考 | 有意识的逻辑分析，速度较慢但更准确 | 复杂问题解决、规划 | <think> 思考过程 </think> 摘要 |
| 最大思考 | 将推理能力发挥到极致 | 探索模型推理能力的边界 | 特殊系统提示词 + <think> 思考过程 </think> 摘要 |
DeepSeek-V4-Pro-Max 与前沿模型对比
| 评测基准（指标） | Opus-4.6 Max | GPT-5.4 xHigh | Gemini-3.1-Pro High | K2.6 Thinking | GLM-5.1 Thinking | DS-V4-Pro Max |
|---|---|---|---|---|---|---|
| 知识与推理 | ||||||
| MMLU-Pro（精确匹配） | 89.1 | 87.5 | 91.0 | 87.1 | 86.0 | 87.5 |
| SimpleQA-Verified（Pass@1） | 46.2 | 45.3 | 75.6 | 36.9 | 38.1 | 57.9 |
| Chinese-SimpleQA（Pass@1） | 76.4 | 76.8 | 85.9 | 75.9 | 75.0 | 84.4 |
| GPQA Diamond (Pass@1) | 91.3 | 93.0 | 94.3 | 90.5 | 86.2 | 90.1 |
| HLE (Pass@1) | 40.0 | 39.8 | 44.4 | 36.4 | 34.7 | 37.7 |
| LiveCodeBench (Pass@1) | 88.8 | - | 91.7 | 89.6 | - | 93.5 |
| Codeforces (Rating) | - | 3168 | 3052 | - | - | 3206 |
| HMMT 2026 Feb (Pass@1) | 96.2 | 97.7 | 94.7 | 92.7 | 89.4 | 95.2 |
| IMOAnswerBench (Pass@1) | 75.3 | 91.4 | 81.0 | 86.0 | 83.8 | 89.8 |
| Apex (Pass@1) | 34.5 | 54.1 | 60.9 | 24.0 | 11.5 | 38.3 |
| Apex Shortlist (Pass@1) | 85.9 | 78.1 | 89.1 | 75.5 | 72.4 | 90.2 |
| 长上下文 | ||||||
| MRCR 1M (MMR) | 92.9 | - | 76.3 | - | - | 83.5 |
| CorpusQA 1M (ACC) | 71.7 | - | 53.8 | - | - | 62.0 |
| 智能体 | ||||||
| Terminal Bench 2.0 (Acc) | 65.4 | 75.1 | 68.5 | 66.7 | 63.5 | 67.9 |
| SWE Verified (Resolved) | 80.8 | - | 80.6 | 80.2 | - | 80.6 |
| SWE Pro (Resolved) | 57.3 | 57.7 | 54.2 | 58.6 | 58.4 | 55.4 |
| SWE Multilingual (Resolved) | 77.5 | - | - | 76.7 | 73.3 | 76.2 |
| BrowseComp (Pass@1) | 83.7 | 82.7 | 85.9 | 83.2 | 79.3 | 83.4 |
| HLE w/ tools (Pass@1) | 53.1 | 52.0 | 51.6 | 54.0 | 50.4 | 48.2 |
| GDPval-AA (Elo) | 1619 | 1674 | 1314 | 1482 | 1535 | 1554 |
| MCPAtlas Public (Pass@1) | 73.8 | 67.2 | 69.2 | 66.6 | 71.8 | 73.6 |
| Toolathlon (Pass@1) | 47.2 | 54.6 | 48.8 | 50.0 | 40.7 | 51.8 |
各模式对比
| 基准测试（指标） | V4-Flash 非思考模式 | V4-Flash High | V4-Flash Max | V4-Pro 非思考模式 | V4-Pro High | V4-Pro Max |
|---|---|---|---|---|---|---|
| 知识与推理 | ||||||
| MMLU-Pro (EM) | 83.0 | 86.4 | 86.2 | 82.9 | 87.1 | 87.5 |
| SimpleQA-Verified (Pass@1) | 23.1 | 28.9 | 34.1 | 45.0 | 46.2 | 57.9 |
| Chinese-SimpleQA (Pass@1) | 71.5 | 73.2 | 78.9 | 75.8 | 77.7 | 84.4 |
| GPQA Diamond (Pass@1) | 71.2 | 87.4 | 88.1 | 72.9 | 89.1 | 90.1 |
| HLE (Pass@1) | 8.1 | 29.4 | 34.8 | 7.7 | 34.5 | 37.7 |
| LiveCodeBench (Pass@1) | 55.2 | 88.4 | 91.6 | 56.8 | 89.8 | 93.5 |
| Codeforces (Rating) | - | 2816 | 3052 | - | 2919 | 3206 |
| HMMT 2026 Feb (Pass@1) | 40.8 | 91.9 | 94.8 | 31.7 | 94.0 | 95.2 |
| IMOAnswerBench (Pass@1) | 41.9 | 85.1 | 88.4 | 35.3 | 88.0 | 89.8 |
| Apex (Pass@1) | 1.0 | 19.1 | 33.0 | 0.4 | 27.4 | 38.3 |
| Apex Shortlist (Pass@1) | 9.3 | 72.1 | 85.7 | 9.2 | 85.5 | 90.2 |
| 长上下文 | ||||||
| MRCR 1M (MMR) | 37.5 | 76.9 | 78.7 | 44.7 | 83.3 | 83.5 |
| CorpusQA 1M (ACC) | 15.5 | 59.3 | 60.5 | 35.6 | 56.5 | 62.0 |
| 智能体 | ||||||
| Terminal Bench 2.0 (Acc) | 49.1 | 56.6 | 56.9 | 59.1 | 63.3 | 67.9 |
| SWE Verified (Resolved) | 73.7 | 78.6 | 79.0 | 73.6 | 79.4 | 80.6 |
| SWE Pro (Resolved) | 49.1 | 52.3 | 52.6 | 52.1 | 54.4 | 55.4 |
| SWE Multilingual (Resolved) | 69.7 | 70.2 | 73.3 | 69.8 | 74.1 | 76.2 |
| BrowseComp (Pass@1) | - | 53.5 | 73.2 | - | 80.4 | 83.4 |
| HLE w/ tools (Pass@1) | - | 40.3 | 45.1 | - | 44.7 | 48.2 |
| MCPAtlas (Pass@1) | 64.0 | 67.4 | 69.0 | 69.4 | 74.2 | 73.6 |
| GDPval-AA (Elo) | - | - | 1395 | - | - | 1554 |
| Toolathlon (Pass@1) | 40.7 | 43.5 | 47.8 | 46.3 | 49.0 | 51.8 |
聊天模板
本次发布不包含 Jinja 格式的聊天模板。我们提供了一个专门的编码文件夹，内含 Python 脚本和测试用例，演示如何将 OpenAI 兼容格式的消息编码为模型的输入字符串，以及如何解析模型的文本输出。完整文档请参考编码文件夹。
简要示例：
from encoding_dsv4 import encode_messages, parse_message_from_completion_text
messages = [
{"role": "user", "content": "hello"},
{"role": "assistant", "content": "Hello! I am DeepSeek.", "reasoning_content": "thinking..."},
{"role": "user", "content": "1+1=?"}
]
prompt = encode_messages(messages, thinking_mode="thinking")
import transformers
tokenizer = transformers.AutoTokenizer.from_pretrained("deepseek-ai/DeepSeek-V4-Pro")
tokens = tokenizer.encode(prompt)
如何在本地运行
关于在本地运行 DeepSeek-V4 的详细说明，包括模型权重转换和交互式聊天演示，请参考推理文件夹。
对于本地部署，我们建议将采样参数设置为 temperature = 1.0，top_p = 1.0。对于 Think Max 推理模式，我们建议将上下文窗口设置为至少 384K 个 token。
许可证
本仓库和模型权重采用 MIT 许可证授权。
引用
@misc{deepseekai2026deepseekv4,
title={DeepSeek-V4: Towards Highly Efficient Million-Token Context Intelligence},
author={DeepSeek-AI},
year={2026},
}
联系方式
如有任何问题，请提交 issue 或通过 service@deepseek.com 联系我们。
2,934,763
862B 参数
I64
F8_E8M0
F8_E4M3
I8
deepseek-ai/DeepSeek-V4-Pro 的模型树
使用 deepseek-ai/DeepSeek-V4-Pro 的 Spaces 100
包含 deepseek-ai/DeepSeek-V4-Pro 的合集
更新于 4 月 24 日
评估结果
- Gsm8k on openai/gsm8k View evaluation results leaderboard 92.6
- SWE Bench Pro on ScaleAI/SWE-bench_Pro View evaluation results leaderboard 55.4
- Diamond on Idavidrein/gpqa View evaluation results leaderboard 90.1
- Chi Bench on actava/chi-bench View evaluation results source leaderboard 14.2 *
- Prior Authorization on actava/chi-bench View evaluation results source leaderboard 10.7 *
- Utilization Management on actava/chi-bench View evaluation results source leaderboard 28 *
- Care Management on actava/chi-bench View evaluation results source leaderboard 4 *
- Mmlu Pro on TIGER-Lab/MMLU-Pro View evaluation results leaderboard 87.5
DeepSeek-V4: Towards Highly Efficient Million-Token Context Intelligence
Introduction
We present a preview version of DeepSeek-V4 series, including two strong Mixture-of-Experts (MoE) language models — DeepSeek-V4-Pro with 1.6T parameters (49B activated) and DeepSeek-V4-Flash with 284B parameters (13B activated) — both supporting a context length of one million tokens.
DeepSeek-V4 series incorporate several key upgrades in architecture and optimization:
- Hybrid Attention Architecture: We design a hybrid attention mechanism combining Compressed Sparse Attention (CSA) and Heavily Compressed Attention (HCA) to dramatically improve long-context efficiency. In the 1M-token context setting, DeepSeek-V4-Pro requires only 27% of single-token inference FLOPs and 10% of KV cache compared with DeepSeek-V3.2.
- Manifold-Constrained Hyper-Connections (mHC): We incorporate mHC to strengthen conventional residual connections, enhancing stability of signal propagation across layers while preserving model expressivity.
- Muon Optimizer: We employ the Muon optimizer for faster convergence and greater training stability.
We pre-train both models on more than 32T diverse and high-quality tokens, followed by a comprehensive post-training pipeline. The post-training features a two-stage paradigm: independent cultivation of domain-specific experts (through SFT and RL with GRPO), followed by unified model consolidation via on-policy distillation, integrating distinct proficiencies across diverse domains into a single model.
DeepSeek-V4-Pro-Max, the maximum reasoning effort mode of DeepSeek-V4-Pro, significantly advances the knowledge capabilities of open-source models, firmly establishing itself as the best open-source model available today. It achieves top-tier performance in coding benchmarks and significantly bridges the gap with leading closed-source models on reasoning and agentic tasks. Meanwhile, DeepSeek-V4-Flash-Max achieves comparable reasoning performance to the Pro version when given a larger thinking budget, though its smaller parameter scale naturally places it slightly behind on pure knowledge tasks and the most complex agentic workflows.
Model Downloads
| Model | #Total Params | #Activated Params | Context Length | Precision | Download |
|---|---|---|---|---|---|
| DeepSeek-V4-Flash-Base | 284B | 13B | 1M | FP8 Mixed | HuggingFace | ModelScope |
| DeepSeek-V4-Flash | 284B | 13B | 1M | FP4 + FP8 Mixed* | HuggingFace | ModelScope |
| DeepSeek-V4-Pro-Base | 1.6T | 49B | 1M | FP8 Mixed | HuggingFace | ModelScope |
| DeepSeek-V4-Pro | 1.6T | 49B | 1M | FP4 + FP8 Mixed* | HuggingFace | ModelScope |
*FP4 + FP8 Mixed: MoE expert parameters use FP4 precision; most other parameters use FP8.
Evaluation Results
Base Model
| Benchmark (Metric) | # Shots | DeepSeek-V3.2-Base | DeepSeek-V4-Flash-Base | DeepSeek-V4-Pro-Base |
|---|---|---|---|---|
| Architecture | - | MoE | MoE | MoE |
| # Activated Params | - | 37B | 13B | 49B |
| # Total Params | - | 671B | 284B | 1.6T |
| World Knowledge | ||||
| AGIEval (EM) | 0-shot | 80.1 | 82.6 | 83.1 |
| MMLU (EM) | 5-shot | 87.8 | 88.7 | 90.1 |
| MMLU-Redux (EM) | 5-shot | 87.5 | 89.4 | 90.8 |
| MMLU-Pro (EM) | 5-shot | 65.5 | 68.3 | 73.5 |
| MMMLU (EM) | 5-shot | 87.9 | 88.8 | 90.3 |
| C-Eval (EM) | 5-shot | 90.4 | 92.1 | 93.1 |
| CMMLU (EM) | 5-shot | 88.9 | 90.4 | 90.8 |
| MultiLoKo (EM) | 5-shot | 38.7 | 42.2 | 51.1 |
| Simple-QA verified (EM) | 25-shot | 28.3 | 30.1 | 55.2 |
| SuperGPQA (EM) | 5-shot | 45.0 | 46.5 | 53.9 |
| FACTS Parametric (EM) | 25-shot | 27.1 | 33.9 | 62.6 |
| TriviaQA (EM) | 5-shot | 83.3 | 82.8 | 85.6 |
| Language & Reasoning | ||||
| BBH (EM) | 3-shot | 87.6 | 86.9 | 87.5 |
| DROP (F1) | 1-shot | 88.2 | 88.6 | 88.7 |
| HellaSwag (EM) | 0-shot | 86.4 | 85.7 | 88.0 |
| WinoGrande (EM) | 0-shot | 78.9 | 79.5 | 81.5 |
| CLUEWSC (EM) | 5-shot | 83.5 | 82.2 | 85.2 |
| Code & Math | ||||
| BigCodeBench (Pass@1) | 3-shot | 63.9 | 56.8 | 59.2 |
| HumanEval (Pass@1) | 0-shot | 62.8 | 69.5 | 76.8 |
| GSM8K (EM) | 8-shot | 91.1 | 90.8 | 92.6 |
| MATH (EM) | 4-shot | 60.5 | 57.4 | 64.5 |
| MGSM (EM) | 8-shot | 81.3 | 85.7 | 84.4 |
| CMath (EM) | 3-shot | 92.6 | 93.6 | 90.9 |
| Long Context | ||||
| LongBench-V2 (EM) | 1-shot | 40.2 | 44.7 | 51.5 |
Instruct Model
DeepSeek-V4-Pro and DeepSeek-V4-Flash both support three reasoning effort modes:
| Reasoning Mode | Characteristics | Typical Use Cases | Response Format |
|---|---|---|---|
| Non-think | Fast, intuitive responses | Routine daily tasks, low-risk decisions | </think> summary |
| Think High | Conscious logical analysis, slower but more accurate | Complex problem-solving, planning | <think> thinking </think> summary |
| Think Max | Push reasoning to its fullest extent | Exploring the boundary of model reasoning capability | Special system prompt + <think> thinking </think> summary |
DeepSeek-V4-Pro-Max vs Frontier Models
| Benchmark (Metric) | Opus-4.6 Max | GPT-5.4 xHigh | Gemini-3.1-Pro High | K2.6 Thinking | GLM-5.1 Thinking | DS-V4-Pro Max |
|---|---|---|---|---|---|---|
| Knowledge & Reasoning | ||||||
| MMLU-Pro (EM) | 89.1 | 87.5 | 91.0 | 87.1 | 86.0 | 87.5 |
| SimpleQA-Verified (Pass@1) | 46.2 | 45.3 | 75.6 | 36.9 | 38.1 | 57.9 |
| Chinese-SimpleQA (Pass@1) | 76.4 | 76.8 | 85.9 | 75.9 | 75.0 | 84.4 |
| GPQA Diamond (Pass@1) | 91.3 | 93.0 | 94.3 | 90.5 | 86.2 | 90.1 |
| HLE (Pass@1) | 40.0 | 39.8 | 44.4 | 36.4 | 34.7 | 37.7 |
| LiveCodeBench (Pass@1) | 88.8 | - | 91.7 | 89.6 | - | 93.5 |
| Codeforces (Rating) | - | 3168 | 3052 | - | - | 3206 |
| HMMT 2026 Feb (Pass@1) | 96.2 | 97.7 | 94.7 | 92.7 | 89.4 | 95.2 |
| IMOAnswerBench (Pass@1) | 75.3 | 91.4 | 81.0 | 86.0 | 83.8 | 89.8 |
| Apex (Pass@1) | 34.5 | 54.1 | 60.9 | 24.0 | 11.5 | 38.3 |
| Apex Shortlist (Pass@1) | 85.9 | 78.1 | 89.1 | 75.5 | 72.4 | 90.2 |
| Long Context | ||||||
| MRCR 1M (MMR) | 92.9 | - | 76.3 | - | - | 83.5 |
| CorpusQA 1M (ACC) | 71.7 | - | 53.8 | - | - | 62.0 |
| Agentic | ||||||
| Terminal Bench 2.0 (Acc) | 65.4 | 75.1 | 68.5 | 66.7 | 63.5 | 67.9 |
| SWE Verified (Resolved) | 80.8 | - | 80.6 | 80.2 | - | 80.6 |
| SWE Pro (Resolved) | 57.3 | 57.7 | 54.2 | 58.6 | 58.4 | 55.4 |
| SWE Multilingual (Resolved) | 77.5 | - | - | 76.7 | 73.3 | 76.2 |
| BrowseComp (Pass@1) | 83.7 | 82.7 | 85.9 | 83.2 | 79.3 | 83.4 |
| HLE w/ tools (Pass@1) | 53.1 | 52.0 | 51.6 | 54.0 | 50.4 | 48.2 |
| GDPval-AA (Elo) | 1619 | 1674 | 1314 | 1482 | 1535 | 1554 |
| MCPAtlas Public (Pass@1) | 73.8 | 67.2 | 69.2 | 66.6 | 71.8 | 73.6 |
| Toolathlon (Pass@1) | 47.2 | 54.6 | 48.8 | 50.0 | 40.7 | 51.8 |
Comparison across Modes
| Benchmark (Metric) | V4-Flash Non-Think | V4-Flash High | V4-Flash Max | V4-Pro Non-Think | V4-Pro High | V4-Pro Max |
|---|---|---|---|---|---|---|
| Knowledge & Reasoning | ||||||
| MMLU-Pro (EM) | 83.0 | 86.4 | 86.2 | 82.9 | 87.1 | 87.5 |
| SimpleQA-Verified (Pass@1) | 23.1 | 28.9 | 34.1 | 45.0 | 46.2 | 57.9 |
| Chinese-SimpleQA (Pass@1) | 71.5 | 73.2 | 78.9 | 75.8 | 77.7 | 84.4 |
| GPQA Diamond (Pass@1) | 71.2 | 87.4 | 88.1 | 72.9 | 89.1 | 90.1 |
| HLE (Pass@1) | 8.1 | 29.4 | 34.8 | 7.7 | 34.5 | 37.7 |
| LiveCodeBench (Pass@1) | 55.2 | 88.4 | 91.6 | 56.8 | 89.8 | 93.5 |
| Codeforces (Rating) | - | 2816 | 3052 | - | 2919 | 3206 |
| HMMT 2026 Feb (Pass@1) | 40.8 | 91.9 | 94.8 | 31.7 | 94.0 | 95.2 |
| IMOAnswerBench (Pass@1) | 41.9 | 85.1 | 88.4 | 35.3 | 88.0 | 89.8 |
| Apex (Pass@1) | 1.0 | 19.1 | 33.0 | 0.4 | 27.4 | 38.3 |
| Apex Shortlist (Pass@1) | 9.3 | 72.1 | 85.7 | 9.2 | 85.5 | 90.2 |
| Long Context | ||||||
| MRCR 1M (MMR) | 37.5 | 76.9 | 78.7 | 44.7 | 83.3 | 83.5 |
| CorpusQA 1M (ACC) | 15.5 | 59.3 | 60.5 | 35.6 | 56.5 | 62.0 |
| Agentic | ||||||
| Terminal Bench 2.0 (Acc) | 49.1 | 56.6 | 56.9 | 59.1 | 63.3 | 67.9 |
| SWE Verified (Resolved) | 73.7 | 78.6 | 79.0 | 73.6 | 79.4 | 80.6 |
| SWE Pro (Resolved) | 49.1 | 52.3 | 52.6 | 52.1 | 54.4 | 55.4 |
| SWE Multilingual (Resolved) | 69.7 | 70.2 | 73.3 | 69.8 | 74.1 | 76.2 |
| BrowseComp (Pass@1) | - | 53.5 | 73.2 | - | 80.4 | 83.4 |
| HLE w/ tools (Pass@1) | - | 40.3 | 45.1 | - | 44.7 | 48.2 |
| MCPAtlas (Pass@1) | 64.0 | 67.4 | 69.0 | 69.4 | 74.2 | 73.6 |
| GDPval-AA (Elo) | - | - | 1395 | - | - | 1554 |
| Toolathlon (Pass@1) | 40.7 | 43.5 | 47.8 | 46.3 | 49.0 | 51.8 |
Chat Template
This release does not include a Jinja-format chat template. Instead, we provide a dedicated encoding folder with Python scripts and test cases demonstrating how to encode messages in OpenAI-compatible format into input strings for the model, and how to parse the model's text output. Please refer to the encoding folder for full documentation.
A brief example:
from encoding_dsv4 import encode_messages, parse_message_from_completion_text
messages = [
{"role": "user", "content": "hello"},
{"role": "assistant", "content": "Hello! I am DeepSeek.", "reasoning_content": "thinking..."},
{"role": "user", "content": "1+1=?"}
]
prompt = encode_messages(messages, thinking_mode="thinking")
import transformers
tokenizer = transformers.AutoTokenizer.from_pretrained("deepseek-ai/DeepSeek-V4-Pro")
tokens = tokenizer.encode(prompt)
How to Run Locally
Please refer to the inference folder for detailed instructions on running DeepSeek-V4 locally, including model weight conversion and interactive chat demos.
For local deployment, we recommend setting the sampling parameters to temperature = 1.0, top_p = 1.0. For the Think Max reasoning mode, we recommend setting the context window to at least 384K tokens.
License
This repository and the model weights are licensed under the MIT License.
Citation
@misc{deepseekai2026deepseekv4,
title={DeepSeek-V4: Towards Highly Efficient Million-Token Context Intelligence},
author={DeepSeek-AI},
year={2026},
}
Contact
If you have any questions, please raise an issue or contact us at service@deepseek.com.
2,934,763
862B params
I64
F8_E8M0
F8_E4M3
I8
Model tree for deepseek-ai/DeepSeek-V4-Pro
Spaces using deepseek-ai/DeepSeek-V4-Pro 100
Collection including deepseek-ai/DeepSeek-V4-Pro
Updated Apr 24
Evaluation results
- Gsm8k on openai/gsm8k View evaluation results leaderboard 92.6
- SWE Bench Pro on ScaleAI/SWE-bench_Pro View evaluation results leaderboard 55.4
- Diamond on Idavidrein/gpqa View evaluation results leaderboard 90.1
- Chi Bench on actava/chi-bench View evaluation results source leaderboard 14.2 *
- Prior Authorization on actava/chi-bench View evaluation results source leaderboard 10.7 *
- Utilization Management on actava/chi-bench View evaluation results source leaderboard 28 *
- Care Management on actava/chi-bench View evaluation results source leaderboard 4 *
- Mmlu Pro on TIGER-Lab/MMLU-Pro View evaluation results leaderboard 87.5