# Model Hub > Give your plugins one place for language models: providers and keys, named roles, cost limits, and a ledger of what every call cost. ## Start - [Model Hub](https://insurgently.com/docs/model-hub/index.md): Set up your AI providers once, and let every plugin use them, with the cost of each call recorded. - [Getting started](https://insurgently.com/docs/model-hub/getting-started.md): Install Model Hub, add a provider and ask your first question. ## How to - [Add a provider](https://insurgently.com/docs/model-hub/add-a-provider.md): Connect OpenRouter, Anthropic, OpenAI, Google Gemini, Azure OpenAI, xAI, Mistral, DeepSeek, Groq, Cerebras, LiteLLM, a local server or any OpenAI-compatible endpoint, and check that it works. - [OpenAI](https://insurgently.com/docs/model-hub/providers/openai.md): Connect OpenAI's own API with a key, and what the Hub does with its models. - [Google Gemini](https://insurgently.com/docs/model-hub/providers/gemini.md): Connect Google's Gemini API with a key, and how its thinking settings map to the Hub's. - [Azure OpenAI](https://insurgently.com/docs/model-hub/providers/azure.md): Connect an Azure OpenAI resource by its name, key and deployments. - [xAI, Mistral, DeepSeek, Groq and Cerebras](https://insurgently.com/docs/model-hub/providers/presets.md): Five hosted providers that speak the OpenAI chat API, each with its own tile, key and base URL, and prices from models.dev. - [llama.cpp](https://insurgently.com/docs/model-hub/providers/llama-cpp.md): Connect the llama.cpp server running on your computer. Free, with no key. - [LiteLLM](https://insurgently.com/docs/model-hub/providers/litellm.md): Connect your own LiteLLM proxy, with prices from models.dev or typed by you, and the cost LiteLLM billed for each call. - [Set up tiers and limits](https://insurgently.com/docs/model-hub/tiers-and-limits.md): Choose which models serve each kind of work, in what order, and how much they may spend. - [Map a plugin's roles](https://insurgently.com/docs/model-hub/roles.md): Decide which tier, or which models, each plugin's jobs run on. - [See what you spent](https://insurgently.com/docs/model-hub/usage-and-costs.md): Read the usage view, and understand listed cost, billed cost and flagged calls. - [Tools and approvals](https://insurgently.com/docs/model-hub/tools.md): What a tool is, who decides whether it runs, and where every call is recorded. - [Use the note chat](https://insurgently.com/docs/model-hub/note-chat.md): Ask about the open note or a selection, research a question on the web, and put the answer into your note. - [Edit a selection](https://insurgently.com/docs/model-hub/selection-commands.md): Rewrite, shorten, fix, continue, translate or summarize selected text, see the change first, and undo it. - [Use Model Hub from Claude Code, Claude Desktop or Codex](https://insurgently.com/docs/model-hub/mcp.md): Switch on the local MCP server so an agent outside Obsidian can use the tools your plugins register and the skills marked for it. - [Add your own skills](https://insurgently.com/docs/model-hub/skills.md): Keep instructions for the model in a vault folder and use them from the note chat and from other plugins. ## Reference - [Settings](https://insurgently.com/docs/model-hub/settings.md): Every setting, page by page, in the order they show. - [Troubleshooting](https://insurgently.com/docs/model-hub/troubleshooting.md): The messages you may see, and how to fix each one. - [Privacy and costs](https://insurgently.com/docs/model-hub/privacy-and-costs.md): What is sent where, what costs money and where your files are kept. ## For plugin developers - [Add Model Hub to your plugin](https://insurgently.com/docs/model-hub/developers.md): Find the Hub, register your roles and make a first call, in about twenty lines. - [The API, call by call](https://insurgently.com/docs/model-hub/developers/api.md): What each Hub call does, what it takes and what comes back. - [Register tools from your plugin](https://insurgently.com/docs/model-hub/developers/tools.md): Offer what your plugin can do as tools, call tools, run several calls under one budget and undo a write. - [Run a tool loop from your plugin](https://insurgently.com/docs/model-hub/developers/agent-runs.md): Start an agent run on one of your roles, end it with a finish tool, follow its progress, stop it and cap its spend. - [Refusal codes](https://insurgently.com/docs/model-hub/developers/errors.md): Every code a Hub error can carry, and what to do about each. - [Use the Hub without Obsidian](https://insurgently.com/docs/model-hub/developers/headless.md): The headless build, for scripts, evals and command-line tools that call the Hub with Obsidian closed. macOS only for keychain keys. - [API versions](https://insurgently.com/docs/model-hub/developers/versions.md): What each version of the API added, and which versions a Hub serves. ## API reference - [Declarations](https://insurgently.com/docs/model-hub/api/model-hub-api.d.ts): every public declaration in one TypeScript file, with comments - [Reference as data](https://insurgently.com/docs/model-hub/api/api.json): each declaration's name, kind, signature, comment and members ## Changes and plans - [Changelog](https://insurgently.com/docs/model-hub/changelog.json): every version's changes, as JSON - [Roadmap](https://insurgently.com/docs/model-hub/roadmap.json): what's coming, as Soon and Later, as JSON - [Everything in one file](https://insurgently.com/docs/model-hub/llms-full.txt)