local intelligence

can it run

Can the Radeon RX 9070 XT (16GB) run gemma3 27B?

NO

No. 16GB of memory can't hold the 22GB working set this model needs. A smaller quant or a bigger machine is required.

~21–31 tok/s

weights 17GB + KV cache 1.94GB at 4k context vs ~16GB usable → offload-partial. 18.9 GB vs 16 GB usable — partial CPU offload, expect large speed loss

computed roofline: 644.6 GB/s × 0.55–0.82 efficiency window / 17.0GB (Q4_K_M) — bands, never points.

explore the full catalog84 machines indexed · live prices · what each one can run

Specs

model file17GB (GGUF Q4_K_M)
minimum memory22GB working set
this machine16GB GDDR6
params27B
licensegemma
price$749

Run it

model files: hugging face ↗

easy run: ollama pull gemma3:27b

runs with: ollama · llama.cpp · koboldcpp · lm studio

Alternatives

Mac mini M4 (32GB) — cheapest machine that runs it ($999)

every model the Radeon RX 9070 XT (16GB) can run — full list

Radeon AI PRO R9700 (32GB) — upgrade path machine

Buy

Newegg ↗ · Amazon ↗ (affiliate)

2026-09-17 · ← all models on the Radeon RX 9070 XT (16GB) · ← full hardware catalog · prices verified at source, may drift