DeepSeek-V4-Pro
Active Parameters
1.6T
Context Length
1M
Modality
Text
Architecture
Mixture of Experts (MoE)
License
MIT
Release Date
24 Apr 2026
Knowledge Cutoff
-
System Requirements
VRAM requirements for different quantization methods and context sizes
1,024 tokens
Consumer
251x RTX 4090
24GB VRAM
Datacenter
58x NVIDIA A100
80GB VRAM
Apple Silicon
56x Apple M3 Max
128GB VRAM
1,000,000 tokens
Consumer
264x RTX 4090
24GB VRAM
Datacenter
61x NVIDIA A100
80GB VRAM
Apple Silicon
59x Apple M3 Max
128GB VRAM
Architecture Diagram
Evaluation Benchmarks
Rank
#22
| Benchmark | Score | Rank |
|---|---|---|
Web Development WebDev Arena | 1462 | 18 |
General Text Text Arena | 1457 | 19 |
Rankings
Overall Rank
#22
Coding Rank
#39
About DeepSeek-V4-Pro
DeepSeek-V4-Pro is DeepSeek's flagship open-source model with 1.6T total parameters and 49B activated per token. Features a novel hybrid CSA+HCA attention mechanism that achieves 1M context with only 27% of the inference FLOPs and 10% of the KV cache compared to DeepSeek-V3.2. In Think Max mode (DeepSeek-V4-Pro-Max), it achieves state-of-the-art open-source results: SWE-Bench Verified 80.6%, SWE-Bench Pro 55.4%, Terminal-Bench 2.0 67.9%, MRCR 1M 83.5%, GPQA Diamond 90.1%, LiveCodeBench 93.5%, and Codeforces Rating 3206. Supports Non-think, Think High, and Think Max reasoning modes. Available via API as deepseek-v4-pro. Released open-source under MIT license on April 24, 2026.
Technical Specifications
Attention
Attention Structure
DeepSeek Sparse Attention
Attention Heads
128
Key-Value Heads
1
Attention Head Dimension
512
Position Embedding
Absolute Position Embedding
RoPE Theta
10,000
Sliding Window Attention
Yes
Sliding Window Size
128
Sliding Window Ratio
-
Linear Attention
-
Linear Attention Ratio
-
Normalization
RMS Normalization
Activation Function
SwigLU
Dimensions
Hidden Dimension Size
7,168
Number of Layers
61
FFN Intermediate Size (Dense)
3,072
Multi-Token Prediction Heads
1
Tokenizer
Vocabulary Size
129,280
Mixture of Experts
Total Expert Parameters
49.0B
Number of Experts
384
Active Experts
6
Shared Experts
1
FFN Intermediate Size (per Expert)
3,072
Dense Layers Before MoE
-
About DeepSeek V4
DeepSeek-V4 is DeepSeek's latest generation of highly efficient Mixture-of-Experts language models, featuring a novel hybrid attention architecture combining Compressed Sparse Attention (CSA) and Heavily Compressed Attention (HCA) that dramatically improves long-context efficiency. Pre-trained on 32T+ tokens with a comprehensive post-training pipeline including domain-specific expert cultivation and unified model consolidation. Both V4-Pro and V4-Flash support 1M context length as standard, with three reasoning effort modes (Non-think, Think High, Think Max). Released open-source under MIT license on April 24, 2026.