Gemma 3 (local)

Google's open-weight multimodal Gemma 3 running locally via Ollama; 4B / 12B sizes with image input and 128K context.

Copy this sentence and send it to any colleague in Teloa:
In Teloa, open Settings · Local models and pull the model for “Gemma 3 (local)” (teloa.model.local.gemma3).

Source: Teloa official
GitHub: Entry file: catalog/models/teloa.model.local.gemma3.json
License: custom
Review: Reviewed by Teloa on 2026-09-26
Compatibility: Needs configuration · Teloa >=0.2.0-alpha.6 · DSH 0.1.7-rc.1
Requires Ollama installed and running on this machine; pulling weights needs network, using them does not.
Permissions and requirements: License has extra terms: The Gemma Terms of Use and Prohibited Use Policy apply; redistribution must include the terms.
Runs locally · Experimental
Access: Ollama
Direct in mainland China
gemma3:4b: 3.3 GB · min RAM 8 GB · recommended 16 GB
gemma3:12b: 8.1 GB · min RAM 16 GB · recommended 32 GB
License tier: Restricted use
License terms: Gemma Terms of Use
Data flow: Weights are pulled by your local Ollama from registry.ollama.ai and checked against the catalog digest; chat stays on your machine and never passes through this site. This site collects no user information.

Entry teloa.model.local.gemma3 · version 1.0.0