DeepSeek V4-Flash — 显存、速度与本地部署
作者： Jakub Rusinowski · 最后更新：
284B-total MoE with 13B active parameters per token, 1M context, MIT license. Even at Q4 this needs roughly 140-160GB depending on quant — realistically a 2-4x RTX 4090/5090 workstation with heavy CPU offload, or a single 80GB+ datacenter GPU. Local inference is experimental: only community llama.cpp forks (e.g. antirez's branch, ~18 tok/s on a 96GB RTX 6000) run it today; no mainline llama.cpp or stable local Ollama build yet. Treat as 'enthusiast workstation, WIP software' rather than a turnkey local model.
DeepSeek V4-Flash 在 Q4 (experimental) 下约需 172 GB 显存——量化权重加框架开销，不含 KV 缓存。在 Apple Silicon 上，这部分来自统一内存。
按量化级别的显存与速度
计算基准：NVIDIA RTX 4090 (24 GB)。仅含权重与开销：该模型架构未公开，因此未计入 KV 缓存。
| 量化 | 显存 | 显存 | 速度（估算） | 适配 |
|---|---|---|---|---|
| Q2_K 2.63 bpw | 94.2 GB | — | 放不下 | |
| Q3_K_M 3.41 bpw | 121.9 GB | — | 放不下 | |
| Q4_K_M 4.83 bpw | 172.3 GB | — | 放不下 | |
| Q5_K_M 5.67 bpw | 202.1 GB | — | 放不下 | |
| Q6_K 6.56 bpw | 233.7 GB | — | 放不下 | |
| Q8_0 8.50 bpw | 302.6 GB | — | 放不下 | |
| F16 16.00 bpw | 568.8 GB | — | 放不下 |
黑色标记 = NVIDIA RTX 4090 (24 GB) 上的可用显存。 估算来自内存带宽屋顶线模型，详见 方法说明页. DeepSeek V4-Flash 显存计算器 →
运行 DeepSeek V4-Flash
目录中能运行 DeepSeek V4-Flash 的最便宜 GPU 是 Apple M2 Ultra (192 GB).
作为亚马逊联盟成员，我们从符合条件的购买中获得收入。云 GPU 链接为推荐链接——我们可能获得佣金，您无需额外付费。
如何运行 DeepSeek V4-Flash
Cloud-hosted on Ollama. Local inference is work-in-progress llama.cpp forks only.
规格
Preview — The model is released, but these specs are thin or rest on a single source. Individual fields may be wrong.
Preview — The model is released, but these specs are thin or rest on a single source. Individual fields may be wrong.
- 参数量
- 284 Billion (13B active)
- 上下文窗口
- 1,000,000
- 架构
- Mixture-of-Experts
- 提供商
- DeepSeek
- 许可证
- MIT
- 规格量化
- Q4 (experimental)
- 系统内存
- 256 GB
- 记录更新于
- 2026-04-24
Commercial use permitted. No usage restrictions beyond attribution.
质量与使用场景
评分由模型作者或独立评测方发布——衡量质量而非吞吐量，并非我们实测。
我的 GPU 能运行 DeepSeek V4-Flash 吗？
- DeepSeek V4-Flash 在 AMD Ryzen AI Max+ 395 上
- DeepSeek V4-Flash 在 Apple M1 Ultra 上
- DeepSeek V4-Flash 在 Apple M2 Max 上
- DeepSeek V4-Flash 在 Apple M2 Ultra 上
- DeepSeek V4-Flash 在 Apple M4 Max 上
- DeepSeek V4-Flash 在 Apple M5 Max 上
- DeepSeek V4-Flash 在 NVIDIA A100 80GB (PCIe) 上
- DeepSeek V4-Flash 在 NVIDIA DGX Spark 上
- DeepSeek V4-Flash 在 NVIDIA H100 80GB (PCIe) 上
- DeepSeek V4-Flash 在 NVIDIA RTX PRO 6000 Blackwell 上
DeepSeek V4 的其他尺寸
DeepSeek V4-Flash — 常见问题
How much VRAM does DeepSeek V4-Flash need?
About 172 GB at Q4 (experimental) — quantized weights plus framework overhead, before any KV cache. The cache grows with context length and is added on top; the table above folds it in. Apple Silicon counts unified memory toward the same figure.
Does DeepSeek V4-Flash run on an RTX 4090 (24 GB)?
No. DeepSeek V4-Flash needs about 172 GB at Q4 (experimental), more than a single RTX 4090's 24 GB. It needs a larger card, several GPUs, or Apple Silicon with enough unified memory — or it runs with part of the weights offloaded to system RAM, which is much slower.
How do I run DeepSeek V4-Flash locally?
Cloud-hosted on Ollama. Local inference is work-in-progress llama.cpp forks only. Running the published tag would send your prompts to a hosted GPU rather than your own machine.
What other sizes does DeepSeek V4 come in?
DeepSeek V4-Flash (172 GB), DeepSeek V4-Pro (967 GB). Every size shares the family's training and licence; the larger ones score higher and need proportionally more memory.