>>
Run gemma-4-12B-it on Lium
Google · Multimodal · 12B parameters · 23.9 GB BF16 weights · released 2026-05-23
It needs about 29 GB of VRAM. The cheapest setup that runs it is 2x RTX 3090 (48 GB), from $0.40 per hour, billed per second.
Other GPUs that run it
How to run it
Rent the pod above, open a terminal over SSH or Jupyter and serve google/gemma-4-12B-it:
vllm serve google/gemma-4-12B-it --tensor-parallel-size 2Claude Code and Codex can do all of it through the Lium CLI.
Frequently asked questions
How much VRAM does gemma-4-12B-it need?
About 23.9 GB for the released BF16 weights, so plan on 29 GB with room for activations and KV cache at a short context. Longer contexts and bigger batches need more.
What GPU should I rent to run gemma-4-12B-it?
2x RTX 3090 (48 GB). On Lium that setup costs from $0.40 per hour, billed per second with no minimum.
For AI agents (CLI install, agent skill, headless signup): llms-full.txt