> ## Documentation Index
> Fetch the complete documentation index at: https://docs.bor-os.io/llms.txt
> Use this file to discover all available pages before exploring further.

# Gemini

# Google Gemini

Use Gemini models in BOR with a Google AI / Gemini API key. Adapter: `server/providers/gemini.js`.

## Setup

1. Get a key from Google AI Studio (linked from onboarding).
2. In [onboarding](../getting-started/onboarding.md), pick **Google Gemini**, paste your key, and choose a model.
3. BOR validates and you're set.

## Choosing a model

Pick a **current** Gemini model from Google's live docs — ids change across the 2.x/3.x lines, so avoid stale ones. A vision-capable model is recommended for the screenshot-driven features.

## Notes

* Gemini supports context caching via `cachedContents`; BOR's prompt-cache layer uses it where available.
* Image blocks and streaming go through the shared multimodal utilities.
* Change the key/model in `providers.gemini` in `config.json`, or ask BOR to switch.

## Troubleshooting

* **API key invalid / permission** — re-check the key and that the Generative Language API is enabled for it.
* **404 model** — the id retired; pick a current Gemini model.
* See [Troubleshooting](../troubleshooting.md).