Yes — with 6.9 GB to spare
Phi-4-mini 3.8B at Q4_K_M fits your M1 · 16 GB entirely in unified memory at 8K context, at an estimated 18 tokens per second. Past 63K the KV cache pushes it over — quantise the cache to q8_0, or step down a quantisation, to go longer.
Fully in unified memory
8K context
Q4_K_M · 2.2 GB
MIT
Released Feb 2025
MIT-licensed, dense, and unusually strong on instruction following for its size.
The VRAM budget
weights 2.2 GB
Weights 2.2 GB
KV cache @ 8K 1.00 GB
Runtime overhead 0.6 GB
Free 6.9 GB of 10.7 GB
Quantisation ladder
| Quant | Weights | Total @ 8K | Max context | Tok/s | Quality | Fit |
|---|---|---|---|---|---|---|
| F16 | 7.2 GB | 8.8 GB | 23K | 5.3 | Reference | Long context |
| Q8_0 | 3.8 GB | 5.4 GB | 50K | 10 | −0.1% ppl | Long context |
| Q6_K | 2.9 GB | 4.5 GB | 57K | 13 | −0.4% ppl | Long context |
| Q5_K_M | 2.5 GB | 4.1 GB | 60K | 15 | −0.8% ppl | Long context |
| Q4_K_M | 2.2 GB | 3.8 GB | 63K | 18 | −1.9% ppl | Recommended |
| Q3_K_M | 1.7 GB | 3.3 GB | 66K | 22 | −5.4% ppl | Long context |
| Q2_K | 1.5 GB | 3.1 GB | 68K | 25 | −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.
How to run it
$ pip install mlx-lm $ mlx_lm.generate --model mlx-community/Phi-4-mini-instruct-4bit \ --max-tokens 512 --prompt "Hello"
Apple's own array framework. The fastest path on Apple Silicon. More on MLX.
01Download is 2.2 GB. Keep it on an SSD — a first load off a spinning disk takes minutes.
02macOS caps what the GPU may wire down at about 10.7 GB of 16 GB. Raising it with iogpu.wired_limit_mb is possible, and risky.
03There is room to go to 63K context on this card.