gptel

gptel is available in any context:

• Interact with LLMs from anywhere in Emacs (any buffer, shell, minibuffer, wherever).
• Carry on multiple independent conversations, one-off, ad hoc interactions or anything in between.

but unobtrusive: it stays out of your way until you need it.

gptel can talk to most LLM services/runners:

• ChatGPT (OpenAI), Claude (Anthropic), Gemini (Google), Llama.cpp and Ollama models,
• but also Perplexity, Kagi, Deepseek, Azure, GitHub Copilot (chat), xAI and AWS Bedrock,
• and nearly every other inference provider or proxy: Openrouter, PrivateGPT, GitHub Models, Groq, Cerebras, Mistral Le Chat, Moonshot, Anyscale, Sambanova, Novita AI, ...

gptel can use LLMs as agents:

• Supports tool-use to equip LLMs with agentic capabilities.
• Supports Model Context Protocol (MCP) integration using mcp.el.

gptel is featureful, it can:

• stream responses, and will never block Emacs. (If you know you know!)
• handle multi-modal input — include images and documents with queries.
• handle "reasoning" content in LLM responses.

Chats are just regular Emacs buffers. This means you can:

• save chats as regular Markdown/Org/Text files and resume them later.
• edit your previous prompts or LLM responses when continuing a conversation. These will be fed back to the model.
• use gptel seamlessly in Org-mode.

gptel is transparent and composable.

• It supports introspection, so you can see exactly what will be sent. Inspect and modify queries before sending them.
• Pause multi-stage requests at an intermediate stage and resume them later.
• Don't like gptel's workflow? gptel is fully scriptable. Use it to create your own for any supported model/backend with a simple API.