DeepSeek-V4-Flash-NVFP4
The NVIDIA DeepSeek-V4-Flash-NVFP4 model is a quantized version of DeepSeek AI's DeepSeek-V4-Flash model, an autoregressive Mixture-of-Experts language model that uses an optimized Transformer architecture with hybrid attention (Compressed Sparse Attention and Heavily Compressed Attention) and Manifold-Constrained Hyper-Connections. For more information, refer to the DeepSeek-V4-Flash model card. The NVIDIA DeepSeek-...
Params
166.7 B
Context
1,048,576
Downloads 30d
203 K
Likes
113
Download history
daily snapshots · 56 days
▲ 113 K in the last 30 days (35.8%)
313 K203 K
Aug 23Sep 2Sep 12Sep 21
1.3 M203 K
Jul 28Aug 15Sep 3Sep 21
Can you run it?
Estimated VRAM at 8K context unless noted. Pick your hardware to see the verdict per quantization.
| File | Quant | Size | Est. VRAM | Verdict on RTX 4090 · 24 GB |
|---|---|---|---|---|
| model.safetensors | u8 | 168.3 GB | 210.6 GB | ❌ Won’t fit |
| model.safetensors (bf16, full) | bf16 + 1024K ctx | 168.3 GB | 3,386.8 GB | ❌ Won’t fit |
Estimate: file size × 1.1 + KV cache at 8K + 0.5 GB overhead. Not a benchmark — how we calculate this.
Run it
copy-paste, exact tags checked against the Hub$ curl -s https://aimodelscomparison.com/api/v1/models/deepseek-v4-flash-nvfp4
{
"hf_id": "nvidia/DeepSeek-V4-Flash-NVFP4",
"params_b": 166.73,
"context_length": 1048576,
"license": { "id": "mit", "commercial": "yes" },
"downloads_30d": 202991,
"vram_estimates": [
{ "quant": "u8", "gb": 210.6 }
],
"updated_at": "2026-07-28T18:03:17Z"
}
Specifications
- Architecture
- DeepseekV4ForCausalLM
- Parameters
- 166.7 B
- Tensor type
- U8
- Context length
- 1,048,576
- Vocabulary
- 129,280
- Layers / heads
- 43 / 64
- Licence
- mit
- First seen on the Hub
- 2026-05-18
- Base model
- DeepSeek-V4-Flash
- Training datasets
- undisclosed
- Added to our catalog
- 2026-07-28
Family
Base model and the most-downloaded derivatives in the catalog.
Compare with any text-generation model