Yes — with 31.3 GB to spare
DeepSeek V4 Flash 284B-A13B at Q4_K_M fits your M3 Ultra · 256 GB entirely in unified memory at 8K context, at an estimated 29 tokens per second. Past 671K the KV cache pushes it over — quantise the cache to q8_0, or step down a quantisation, to go longer.
The V4 that 128 GB machines can actually run at Q3. Cache is modelled as a 576-wide latent; V4 compresses it further at long context, so this is conservative.
The VRAM budget
Quantisation ladder
| Quant | Weights | Total @ 8K | Max context | Tok/s | Quality | Fit |
|---|---|---|---|---|---|---|
| Q8_0 | 281.0 GB | 282.0 GB | — | ~16 | −0.1% ppl | 90.0 GB over |
| Q6_K | 216.9 GB | 217.9 GB | — | ~21 | −0.4% ppl | 25.9 GB over |
| Q5_K_M | 187.5 GB | 188.4 GB | 83K | 24 | −0.8% ppl | Long context |
| Q4_K_M | 159.7 GB | 160.7 GB | 671K | 29 | −1.9% ppl | Recommended |
| Q3_K_M | 129.3 GB | 130.3 GB | 1024K | 35 | −5.4% ppl | Long context |
| Q2_K | 110.8 GB | 111.7 GB | 1024K | 41 | −15% ppl | Long context |
Quality is the published perplexity delta against f16 weights. Max context assumes an f16 KV cache; q8_0 roughly doubles it. This model uses multi-head latent attention, so its cache is a compressed latent rather than full K and V.
How to run it
$ pip install mlx-lm $ mlx_lm.generate --model mlx-community/DeepSeek-V4-Flash-0731-4bit \ --max-tokens 512 --prompt "Hello"
Apple's own array framework. The fastest path on Apple Silicon. More on MLX.