Yes, just — 2.0 GB spare

Gemma 4 26B-A4B at Q4_K_M fits your M2 · 24 GB entirely in unified memory at 8K context, at an estimated 12 tokens per second. Past 58K 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 · 14.9 GB Apache 2.0 Released 2 Apr 2026 Vision

Mixture of experts with 3.8B active. Slower to think than Qwen3.6 35B-A3B, faster to answer, and it sees images.

What hardware do I need for Gemma 4 26B-A4B? →

The VRAM budget

weights 14.9 GB

Weights 14.9 GB KV cache @ 8K 0.51 GB Runtime overhead 0.6 GB Free 2.0 GB of 18.0 GB

Quantisation ladder

Quant	Weights	Total @ 8K	Max context	Tok/s	Quality	Fit
Q8_0	26.2 GB	27.3 GB	—	~6.9	−0.1% ppl	9.3 GB over
Q6_K	20.2 GB	21.3 GB	—	~9.0	−0.4% ppl	3.3 GB over
Q5_K_M	17.5 GB	18.6 GB	—	~10	−0.8% ppl	0.6 GB over
Q4_K_M	14.9 GB	16.0 GB	58K	12	−1.9% ppl	Recommended
Q3_K_M	12.1 GB	13.2 GB	131K	15	−5.4% ppl	Long context
Q2_K	10.3 GB	11.4 GB	175K	18	−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, 5 of 30 layers global), which is why its cache barely grows with context.

How to run it

Ollama llama.cpp LM Studio MLX

terminal

$ pip install mlx-lm
$ mlx_lm.generate --model mlx-community/gemma-4-26B-A4B-it-4bit \
    --max-tokens 512 --prompt "Hello"

Apple's own array framework. The fastest path on Apple Silicon. More on MLX.

01Download is 14.9 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 18.0 GB of 24 GB. Raising it with iogpu.wired_limit_mb is possible, and risky.

03Only 2.0 GB is spare, so a long system prompt can still push it over. Its real ceiling here is 58K context.

See all models for this rig Compare with another model