Smaller alternatives
Models that need less memory than DeepSeek V4 Flash 0731 and score about as well or better.
No scored model is both smaller and about as intelligent as DeepSeek V4 Flash 0731.
DeepSeek V4 Flash 0731
MoE
DeepSeek · 304.2B (13B active) · Mixture of experts
DeepSeek V4 Flash 0731 is a 304.2B model from DeepSeek with a 1M-token context. At Q4_K_M with an 8K context it needs about 161 GB; 203 GB leaves room for longer chats.
Mixture of experts
Parameters: 304.2B · Active: 13B
All 304.2B parameters load into memory, but only 13B work on each token, so it runs at the speed of a much smaller model.
Quantization options
| Quant | Bits | Memory | Quality | On your hardware |
|---|---|---|---|---|
| Q2_K | 3.16 | — | Noticeably worse | … |
| Q3_K_M | 4 | — | Some loss | … |
| Q4_K_M | 4.89 | — | Good | … |
| Q5_K_M | 5.7 | — | Good | … |
| Q6_K | 6.56 | — | Near original | … |
| Q8_0 | 8.5 | — | Near original | … |
| F16 | 16 | — | Original | … |
Memory by context length
At Q4_K_M. The context cache grows with every token the model keeps in mind.
| Context | Context cache | Total |
|---|---|---|
| 4K | 0.3 GB | 160.7 GB |
| 8K | 0.7 GB | 161 GB |
| 32K | 2.7 GB | 163 GB |
| 128K | 10.8 GB | 171.1 GB |
| 256K | 21.5 GB | 181.8 GB |
Next steps
Compare it with
Scored models of a similar size, side by side on your device.
Least hardware that runs it well
Desktop devices with the least memory that give DeepSeek V4 Flash 0731 grade A or S at Q4_K_M.
Can I run DeepSeek V4 Flash 0731 locally?
Can I run DeepSeek V4 Flash 0731 locally?
Yes, if your GPU or Mac has about 161 GB free for it at Q4_K_M. Open this page on that computer to see the grade for your exact hardware, then start it with llama.cpp, Ollama or LM Studio.
How much memory does DeepSeek V4 Flash 0731 need?
About 161 GB at Q4_K_M with an 8K context and 282.4 GB at Q8_0. Lower quantizations fit smaller cards with some loss in quality; longer contexts add to the total.