How to Use Gemma 3 4B on Composite
At a glance
| Model ID | google/gemma-3-4b-it |
|---|---|
| Free ID | google/gemma-3-4b-it:free |
| Made by | |
| Context window | 131,072 tokens |
| Longest reply | 117,964 tokens |
| Input | Text, images |
| Reasoning | No |
| Tool calling | No |
Gemma 3 4B is the instruction-tuned 4B model from Google's Gemma 3 family. Gemma 3 added image input to the line, handles a 128k-token window, and Google lists support for over 140 languages along with better math, reasoning and chat than earlier Gemma releases. At 4B parameters, it is the smallest free model with a published size.
Where a small model makes sense
Four billion parameters is small next to the mixture-of-experts models on the list, and it shows in long, layered scenes. Gemma 3 4B is better used for short jobs: translating a message, describing an image, rewriting a paragraph in a different tone, or writing a summary of a chat you paste in. Its language coverage makes translation a natural fit.
If you want a longer story on the free list, compare it against Mistral Nemo or one of the larger Flash models before settling.
Images
Gemma 3 4B reads images. Attach a picture in apps that support it, or send an image_url content part through the API. Replies are text only.
Settings
It accepts the usual chat samplers, including min P and repetition penalty, plus stop sequences and a JSON response format. It does not support tool calling or reasoning effort, so leave tools out of the request and do not add a bracket suffix such as [high] to the model ID.
Connect it to your chat app
- Create an API key on the API Keys page.
- In SillyTavern, ChubAI, Agnai or RisuAI, choose an OpenAI-compatible provider and set the base URL to
https://composite.lucidity.sh/v1. JanitorAI wants the full URL:https://composite.lucidity.sh/v1/chat/completions. - Paste the key, then type
google/gemma-3-4b-it:freeas the model name. - Send a short message to check the connection before you load a long chat.
Step-by-step setup for each app is in the JanitorAI and SillyTavern guides, and the docs hub covers the others.
From code, it is a normal OpenAI-compatible request:
curl https://composite.lucidity.sh/v1/chat/completions \
-H "Authorization: Bearer YOUR_API_KEY" \
-H "Content-Type: application/json" \
-d '{"model": "google/gemma-3-4b-it:free", "messages": [{"role": "user", "content": "Hello"}]}'Free access (as of October 21, 2026)
On October 21, 2026, this model is on Composite's free list as google/gemma-3-4b-it:free. The plain google/gemma-3-4b-it stays paid. Every account shares 80 requests a day across all of the free models, separate from Synth's count, and the count resets at midnight UTC. Credits and plans do not raise it.
The free list changes. Check the free models on the Models page for the current set, and the features guide for the rules. A :free id that is no longer on the list returns a model-not-found error and charges nothing. When the 80 requests are used up, requests stop with an error until the reset; Composite never charges your credits to continue. You can switch to google/gemma-3-4b-it on credits or a plan, or to Synth.