Yes — with 46.4 GB to spare

Gemma 4 26B-A4B at Q4_K_M fits your Instinct MI210 entirely on the GPU at 8K context, at an estimated 178 tokens per second. There is room for its full 256K window.

Fully on GPU 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 46.4 GB of 62.4 GB

Quantisation ladder

Quant	Weights	Total @ 8K	Max context	Tok/s	Quality	Fit
Q8_0	26.2 GB	27.3 GB	256K	101	−0.1% ppl	Long context
Q6_K	20.2 GB	21.3 GB	256K	131	−0.4% ppl	Long context
Q5_K_M	17.5 GB	18.6 GB	256K	152	−0.8% ppl	Long context
Q4_K_M	14.9 GB	16.0 GB	256K	178	−1.9% ppl	Recommended
Q3_K_M	12.1 GB	13.2 GB	256K	220	−5.4% ppl	Long context
Q2_K	10.3 GB	11.4 GB	256K	257	−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 vLLM

terminal

$ ollama pull gemma4:26b
$ OLLAMA_CONTEXT_LENGTH=8192 \
    ollama run gemma4:26b

The default. One binary, a model registry, an OpenAI-compatible port. More on Ollama.

01Download is 14.9 GB. Keep it on an SSD — a first load off a spinning disk takes minutes.

02Close anything else holding VRAM. A browser with hardware acceleration can sit on 1–2 GB.

03There is room to go to the model's full 256K context on this card.

See all models for this rig Compare with another model