---
title: Gemma 4 E4B ZeroGPU
emoji: 🔥
colorFrom: indigo
colorTo: purple
sdk: gradio
sdk_version: 6.15.2
app_file: app.py
pinned: false
short_description: OpenAI-compatible Gemma 4 E4B chat endpoint on ZeroGPU
---

# Gemma 4 E4B — ZeroGPU OpenAI-compatible endpoint

On-demand (`@spaces.GPU`) Gemma 4 E4B chat server exposing an
OpenAI-shaped `POST /v1/chat/completions` route. Sized for bursty,
one-off generation (companion / script generation).

- Model: `OBLITERATUS/gemma-4-E4B-it-OBLITERATED` (overridable via `MODEL_ID`).
- Endpoint base: `https://<space-host>/v1`
- Routes: `POST /v1/chat/completions`, `GET /health`, Gradio UI at `/ui`.
- Optional bearer auth via the `ENDPOINT_API_KEY` secret.

Request/response follow the OpenAI chat-completions shape:
`{ "model", "messages": [{role, content}], "temperature", "max_tokens" }`
→ `{ "choices": [{ "message": { "content" } }] }`.