Freedom251 › AI chatbots and assistants
OpenThena

What OpenThena is
OpenThena is a free, MIT-licensed AI assistant that streams responses and keeps chat history in a browser. Xiaomi MiMo v2.5 is the default model, while configuration can select another provider. A LiteLLM-style gateway exposes an OpenAI-format model interface, and an optional LiteLLM proxy can centralize keys, budgets, and providers. Responses arrive token by token through the AI SDK data-stream protocol. OpenThena says it has no database, accounts, message-content server logs, analytics, fingerprinting, or advertising cookies. Conversation history is stored in browser localStorage and can be erased by clearing a conversation or browser data. Prompts are sent to the configured upstream provider, whose terms govern handling. The Next.js app can run wherever Node.js runs, with one-click Vercel deployment or full-stack self-hosting. Local setup requires Node.js 18+ and a Xiaomi MiMo API key.
Who it suits
It suits users who want a self-hostable, open-source assistant with configurable model providers. Developers can run it on Node.js, deploy through Vercel, or connect OpenAI-compatible endpoints.
What it does well
- Free and MIT licensed
- Token-by-token response streaming
- Configurable providers and OpenAI-format gateway
- No accounts, analytics, or advertising cookies
- Supports Vercel deployment and self-hosting
What to know first
- Model-provider charges may apply
- Prompts follow the upstream provider’s terms
- Local setup requires Node.js 18+ and an API key
- Chat API accepts at most 50 messages per request
Plans
| Plan | Billed and limits | Price |
|---|---|---|
| OpenThena | Free; model provider charges may apply · OpenThena itself is free · pay only for the model provider you use | Free |
As openthena.com lists them, read 11 Oct 2026. The maker's own page has the final word.
Verdict
OpenThena focuses on configurable, self-hosted AI chat with minimal stated application data collection. Users remain responsible for provider costs and for how the selected model service handles prompts.
Facts from OpenThena's own pages
- API limit
- The chat API accepts at most 50 messages per request. · openthena.com, 5 Oct 2026
- Chat storage
- Chat history is saved in the browser's localStorage and can be erased by clearing a conversation or browser data. · openthena.com, 11 Oct 2026
- Cost limit
- OpenThena itself is free; users pay only for the model provider they choose. · openthena.com, 11 Oct 2026
- Default model
- It uses Xiaomi MiMo (mimo-v2.5-pro) by default, and the model can be changed through an environment variable. · openthena.com, 11 Oct 2026
- Deployment
- The documentation describes one-click Vercel deployment and self-hosting. · openthena.com, 11 Oct 2026
- Free plan
- Yes · openthena.com, 21 Sept 2026
- Gateway
- Its LiteLLM-style gateway provides an OpenAI-format interface to models. · openthena.com, 11 Oct 2026
- History storage
- Chat history is stored in browser localStorage and can be erased by clearing a conversation or browser data. · openthena.com, 5 Oct 2026
- License
- OpenThena is MIT licensed and described as auditable and open source. · openthena.com, 11 Oct 2026
- Local setup
- The quickstart requires Node.js 18+ and a Xiaomi MiMo API key to run locally. · openthena.com, 5 Oct 2026
- Model
- Xiaomi MiMo v2.5 is the default model, and the provider can be changed through configuration. · openthena.com, 5 Oct 2026
- Model compatibility
- The FAQ says any OpenAI-compatible endpoint works, including a self-hosted LiteLLM proxy or a local model. · openthena.com, 11 Oct 2026
- Privacy
- The maker says it has no database, accounts, message-content server logs, analytics, fingerprinting, or advertising cookies. · openthena.com, 11 Oct 2026
- Product
- OpenThena is an open-source AI assistant with a streaming chat experience. · openthena.com, 11 Oct 2026
- Provider data
- Prompts are sent to the configured upstream model provider, whose data handling is governed by that provider’s terms. · openthena.com, 5 Oct 2026
- Provider data handling
- Prompts are sent to the configured upstream model provider, whose data handling is governed by that provider's terms. · openthena.com, 11 Oct 2026
- Self-hosting
- OpenThena is a Next.js app that can run anywhere Node runs. · openthena.com, 11 Oct 2026
- Streaming
- Responses stream token by token. · openthena.com, 11 Oct 2026
Where OpenThena stands on Freedom251
- AI chatbots and assistants#51 of 65