Yes — with 7.0 GB to spare
Gemma 3 27B at Q4_K_M fits your M2 Pro · 32 GB entirely in unified memory at 8K context, at an estimated 7.3 tokens per second. Past 97K 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 · 15.4 GB
Gemma Terms of Use
Released Mar 2025
Vision
The 2025 single-GPU generalist with vision. Its Gemma-licence terms are the reason to prefer Gemma 4 now.
The VRAM budget
weights 15.4 GB
Weights 15.4 GB
KV cache @ 8K 1.03 GB
Runtime overhead 0.6 GB
Free 7.0 GB of 24.0 GB
Quantisation ladder
| Quant | Weights | Total @ 8K | Max context | Tok/s | Quality | Fit |
|---|---|---|---|---|---|---|
| Q8_0 | 27.1 GB | 28.7 GB | — | ~4.1 | −0.1% ppl | 4.7 GB over |
| Q6_K | 20.9 GB | 22.6 GB | 26K | 5.3 | −0.4% ppl | Long context |
| Q5_K_M | 18.1 GB | 19.7 GB | 62K | 6.2 | −0.8% ppl | Long context |
| Q4_K_M | 15.4 GB | 17.0 GB | 97K | 7.3 | −1.9% ppl | Recommended |
| Q3_K_M | 12.5 GB | 14.1 GB | 128K | 9.0 | −5.4% ppl | Long context |
| Q2_K | 10.7 GB | 12.3 GB | 128K | 10 | −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 (1024 tokens, 1 global in 6), which is why its cache barely grows with context.
How to run it
$ pip install mlx-lm $ mlx_lm.generate --model mlx-community/gemma-3-27b-it-4bit \ --max-tokens 512 --prompt "Hello"
Apple's own array framework. The fastest path on Apple Silicon. More on MLX.
01Download is 15.4 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 97K context on this card.