Yes — with 18.8 GB to spare
Gemma 4 E4B at Q4_K_M fits your M2 Pro · 32 GB entirely in unified memory at 8K context, at an estimated 25 tokens per second. There is room for its full 128K window.
Fully in unified memory
8K context
Q4_K_M · 4.5 GB
Apache 2.0
Released 2 Apr 2026
Vision
The laptop Gemma. 4.5B effective, 8B on disk; a single KV head per window layer keeps its cache tiny.
The VRAM budget
weights 4.5 GB
Weights 4.5 GB
KV cache @ 8K 0.14 GB
Runtime overhead 0.6 GB
Free 18.8 GB of 24.0 GB
Quantisation ladder
| Quant | Weights | Total @ 8K | Max context | Tok/s | Quality | Fit |
|---|---|---|---|---|---|---|
| F16 | 14.9 GB | 15.6 GB | 128K | 7.5 | Reference | Long context |
| Q8_0 | 7.9 GB | 8.7 GB | 128K | 14 | −0.1% ppl | Long context |
| Q6_K | 6.1 GB | 6.9 GB | 128K | 18 | −0.4% ppl | Long context |
| Q5_K_M | 5.3 GB | 6.0 GB | 128K | 21 | −0.8% ppl | Long context |
| Q4_K_M | 4.5 GB | 5.2 GB | 128K | 25 | −1.9% ppl | Recommended |
| Q3_K_M | 3.6 GB | 4.4 GB | 128K | 31 | −5.4% ppl | Long context |
| Q2_K | 3.1 GB | 3.9 GB | 128K | 36 | −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 interleaves sliding-window layers (512 tokens, 7 of 42 layers global), which is why its cache barely grows with context.
How to run it
$ pip install mlx-lm $ mlx_lm.generate --model mlx-community/gemma-4-E4B-it-4bit \ --max-tokens 512 --prompt "Hello"
Apple's own array framework. The fastest path on Apple Silicon. More on MLX.
01Download is 4.5 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 24.0 GB of 32 GB. Raising it with iogpu.wired_limit_mb is possible, and risky.
03There is room to go to the model's full 128K context on this card.