> ## Documentation Index > Fetch the complete documentation index at: https://docs.snaply.ai/llms.txt > Use this file to discover all available pages before exploring further. # Cloud Models > Use OpenAI, Anthropic, Google Gemini, OpenRouter, xAI, Groq, Mistral, or an OpenAI-compatible endpoint with Snaply. **Bring the model setup you already like into Snaply.** The selected model powers [Writing Assistant](/writing-assistant/overview), meeting titles, AI meeting notes, and [Meeting Notes Ask AI](/meeting-notes/ask-ai). Use the local model when you want the most private, offline-friendly default. Switch to a cloud model when you want a specific provider, a larger model, stronger reasoning, or the same model your team already uses elsewhere. This is the right page if you are looking for broader model support, OpenRouter, Claude/OpenAI/Gemini-style presets, Mistral, or a local OpenAI-compatible setup with Ollama, LM Studio, or LiteLLM. ## Supported Providers Snaply includes built-in provider setup for: * **OpenAI** * **Anthropic** * **Google Gemini** * **OpenRouter** * **xAI** * **Groq** * **Mistral** * **OpenAI-compatible endpoint** The catalog contains hundreds of models. Snaply keeps the first choice simple by showing recommended models first, then gives you search when you want to pick a specific model. ## Common Requested Setups | What you want | Where to start | | --------------------------------------------------------- | -------------------------------------------------------------------- | | OpenAI, Claude, Gemini, Mistral, Groq, xAI, or OpenRouter | Pick the provider in **Advanced model setup** | | A local Ollama model | Use **OpenAI-compatible endpoint** with `http://localhost:11434/v1` | | LM Studio's local server | Use **OpenAI-compatible endpoint** with `http://localhost:1234/v1` | | LiteLLM, a private proxy, or a company gateway | Use **OpenAI-compatible endpoint** with that server's `/v1` base URL | | One API key for many model families | Use **OpenRouter** | ## The Simple Setup 1. Open **Settings -> Writing Assistant -> AI Models**. 2. Go to **Advanced model setup**. 3. Pick the provider you want. 4. Choose a recommended model, or search the full catalog for a specific one. 5. Click **Add API key** and paste the provider key. 6. Select **Use** on the model you want. That is it. From then on, Writing Assistant, meeting titles, AI notes, and Ask AI use that model. Your provider account still controls pricing, rate limits, billing, and data handling for that provider. ## OpenAI-Compatible Endpoint Choose **OpenAI-compatible endpoint** when you want Snaply to talk to something custom: [Ollama](https://docs.ollama.com/api/openai-compatibility), [LM Studio](https://lmstudio.ai/docs/developer/openai-compat), LiteLLM, a local model server, private proxy, company gateway, or hosted provider with an OpenAI-style API. Snaply asks for: * **Base URL**: the endpoint root, for example `http://localhost:11434/v1` * **Model name**: the exact model identifier your endpoint expects * **API key**: optional, only needed if the endpoint requires a bearer token After you save it, the endpoint behaves like any other AI model in Snaply. ## Ollama, LM Studio, And LiteLLM This is the easiest way to bring local models you already run elsewhere into Snaply. | Tool | Base URL to try | Model name | | ------------------------------------------- | ---------------------------- | ------------------------------------------------------------------- | | Ollama | `http://localhost:11434/v1` | The model you pulled in Ollama, such as `llama3.2` or `llama3.1:8b` | | LM Studio | `http://localhost:1234/v1` | The model identifier shown by LM Studio's local server | | LiteLLM or another OpenAI-compatible server | Your server's `/v1` endpoint | The model name that server expects | For local tools like Ollama and LM Studio, you can usually leave **API key** empty unless you turned on authentication yourself. If you type `http://localhost:11434` or `http://localhost:1234`, Snaply can suggest the matching `/v1` URL. ## API Key Storage Cloud API keys are stored in Snaply's encrypted app vault on your Mac. From the API key dialog, you can: * Add a key for each provider * See which providers have a key saved * Reveal a saved key when needed * Remove a key * Load previously saved keys from older Snaply builds OpenAI-compatible endpoints can be used with or without a key, depending on how your endpoint is configured. ## What Gets Sent To A Cloud Model When a cloud model is selected, Snaply sends only the text needed for the request to that provider. That can include: * Selected text for Writing Assistant edits * The prompt and chat history for Writing Assistant chat * Meeting transcript content for generated titles and notes * Meeting transcript and notes context for Ask AI Meeting recordings and transcripts are still stored locally by Snaply. The important distinction is generation: if the selected model is cloud, the text needed for that generation goes to the selected provider. For OpenAI-compatible endpoints, the text goes to the endpoint URL you configured. If that endpoint is Ollama or LM Studio running on `localhost`, the request stays on your machine. If it is LiteLLM, a hosted gateway, or a company proxy, the request goes there. ## When To Use Each Option | Use case | Good choice | | ------------------------------------------------------------------ | ----------------------------------- | | Private offline rewriting | Local Daily Assistant model | | Strong general writing and reasoning | OpenAI, Anthropic, or Google Gemini | | Access to many model families from one key | OpenRouter | | Fast hosted inference | Groq | | Mistral or Codestral workflows | Mistral | | Grok models | xAI | | Ollama, LM Studio, LiteLLM, local server, proxy, or custom gateway | OpenAI-compatible endpoint | ## Troubleshooting If a cloud model cannot run, Snaply shows an error card with the next step. Most issues come down to one of these: * Add the missing API key * Choose another model from the same provider * Check quota, billing, or rate limits in the provider account * For OpenAI-compatible endpoints, confirm the base URL, model name, and bearer token ## Related Pages * [AI Models overview](/writing-assistant/models) * [Writing Assistant modes](/writing-assistant/modes) * [Meeting Notes Ask AI](/meeting-notes/ask-ai)