# Google Gemini

URL: https://docs.hanzo.ai/docs/guides/migrate/gemini

> Gemini models are in the catalogue as google/gemini-*, served on the OpenAI-compatible routes at api.hanzo.ai/v1. Gemini's own generateContent wire is not served.

Google's Gemini API is its own wire, `models/{model}:generateContent`.
Hanzo serves Gemini models on the OpenAI-compatible routes instead:
`POST https://api.hanzo.ai/v1/chat/completions` with `model:
"google/gemini-2.5-flash"` or any other `google/…` id the catalogue lists. Moving
over means switching from the Gemini SDK to an OpenAI-compatible client, after
which every other provider's models are one name away.

## Start here

```bash
curl -sS https://api.hanzo.ai/v1/chat/completions \
  -H "Authorization: Bearer $HANZO_API_KEY" \
  -H 'Content-Type: application/json' \
  -d '{"model":"google/gemini-2.5-flash","messages":[{"role":"user","content":"summarise Q3"}]}'

curl -sS https://api.hanzo.ai/v1/models | jq -r '.data[].id' | grep '^google/'
```

## Gemini CLI

Keep Gemini CLI on your Google account. `hanzo --gemini` attaches the Hanzo
toolset for the run, so it can hand work to Enso and Kai. It speaks only Google's
API, so `--cloud` does not apply to it. [Gemini CLI with Hanzo →](https://docs.hanzo.ai/docs/guides/coding/agents#gemini-cli)

## What does not carry

**`generateContent` is not served.** A client built on the Gemini wire
(`v1beta/models/…:generateContent`) needs an OpenAI-compatible client to reach
api.hanzo.ai.

**Safety settings and Gemini-only request fields stay with Google.** The
OpenAI-compatible body carries messages, tools and sampling settings.