Agentic LLM

Google's Gemma 4, locally or via API.

Gemma 4 is the next generation of Google's open-weights model family. Run it through Ollama on your own hardware, or via Google's hosted API when you need higher-tier capability without local GPU.

Choose Your Plan Back to overview

Google's Gemma 4 open-weights model, runnable locally via Ollama or via Google's hosted API.

What it does

Gemma 4 at a glance.

Open weights, your hardware

Download Gemma 4 weights through Ollama and run on local GPU. Same speed-of-thought as Llama 3 with a slightly different reasoning shape.

Hosted API fallback

If your local hardware can't handle the larger Gemma 4 variants, Google's hosted API ships the same weights at faster token throughput. Bring your own key.

Multimodal

Gemma 4 reads images natively. Useful for Agent Eyes-paired flows, OCR jobs, and any task where vision belongs in the prompt.

Bring it where you need it

Configure as the brain for specific tab agents, specific tasks, or as a global default. Each agent in VP3 can pick its own brain.

Add Gemma 4 to your agent.

Every VP3 plan unlocks the full c:// drive catalog. Pick a plan, install the home server, flip the app on.

Choose Your Plan