作者： Jakub Rusinowski · 最后更新： 2026年7月21日
Rank #1 open-weight model on every major benchmark. DeepSeek V4 Pro uses a massive 1.6 trillion parameter Mixture-of-Experts architecture with only 49 billion parameters active per token — giving frontier-tier intelligence at a fraction of the compute cost. Dominates coding, math, reasoning, and multilingual tasks.
| DeepSeek V4 Pro (Q4_K_M) | Min 967 GB VRAM · Q4_K_M · 128,000 ctx · ollama run deepseek-v4-pro:latest |
| DeepSeek V4 Pro (API) | Min 0 GB VRAM · None (cloud) · 128,000 ctx · |
The cheapest GPU that runs DeepSeek V4 Pro (Legacy Listing — Unverified Context/License) locally (min 4 GB VRAM) is the Intel Arc B570 (10 GB).
Install Ollama then run: ollama run deepseek-v4-pro:latest
Minimum VRAM: 4 GB. For best results use Q4_K_M quantization.
DeepSeek V4 Pro (Legacy Listing — Unverified Context/License) needs about 4 GB VRAM at Q4_K_M quantization for its smallest variant. Variants: DeepSeek V4 Pro (Q4_K_M) (967 GB, Q4_K_M); DeepSeek V4 Pro (API) (0 GB, None (cloud)). On Apple Silicon, unified memory counts toward this requirement.
Yes — DeepSeek V4 Pro (Legacy Listing — Unverified Context/License) runs on an RTX 4090 (24 GB) and other 24 GB cards such as the RTX 3090. Smaller variants also fit comfortably on 8–16 GB GPUs at Q4_K_M.
Q4_K_M is the best balance of quality and VRAM for DeepSeek V4 Pro (Legacy Listing — Unverified Context/License) in most cases. Choose Q8_0 for near-lossless quality if you have spare VRAM, or smaller quants (Q3/Q2) only when memory is tight.
Install Ollama, then run: ollama run deepseek-v4-pro:latest. This downloads DeepSeek V4 Pro (Legacy Listing — Unverified Context/License) and starts a local, OpenAI-compatible endpoint — no internet connection is needed after the initial download.