Query multiple AI models at once and compare their answers side by side using the Beam feature.
Self-host your own private AI chat workspace using Docker so your conversations never leave your server.
Import a PDF document into a conversation and ask questions about its contents.
Use the AI Inspector mode to see exactly what is sent to and received from any AI model.
Requires your own AI provider API keys (e.g. OpenAI, Anthropic) to access models, no keys are provided by the app.
Big-AGI is an open-source AI chat workspace that lets you talk to many different AI models from a single interface. Instead of opening separate websites for different AI services, you connect your own API keys and use one app to access over 500 models across 20+ providers, including OpenAI, Anthropic, Google Gemini, DeepSeek, and others. The headline feature is called Beam. When you ask a question, Beam sends it to multiple AI models at the same time and shows you all the answers side by side. You can then pick the best response or merge responses together. The idea is that running several models in parallel and comparing their outputs catches mistakes that any single model might make on its own. This is particularly useful for technical or high-stakes questions where you want more than one perspective before acting on an answer. The app includes other features you would expect from a chat tool: custom AI personas you can define, text-to-image generation, voice input, code highlighting, the ability to run code in the browser, and importing PDF documents into conversations. There is also an AI Inspector mode that lets you see exactly what was sent to and received from the AI, which is useful if you care about privacy or want to debug what is happening under the hood. You can use it for free on the hosted website at big-agi.com with your own API keys. A Pro tier at around $11 per month adds sync across devices and more storage. Alternatively, you can self-host the entire application using Docker on your own server, or deploy it to a cloud hosting provider. The self-hosted version gives you maximum privacy since your conversations stay on your own infrastructure. The project is independently funded, meaning there is no venture capital backing. Revenue from Pro subscriptions pays for ongoing development. It is released under the MIT open-source license, so you are free to inspect, modify, and redistribute the code.
← enricoros on gitmyhub — every repo by this author, as a profile.
Verify against the repo before relying on details.