Inference Bridge av Sam Samskies

Inference Bridge gives websites a browser API for streaming AI chat, tools, hosted web search, and supported images. You choose a provider and model in Options, then approve each website or save a per-site grant. Bring your own API key for supported cloud providers, use local Ollama, or add an OpenAI-compatible server. Keys remain inside the add-on; page scripts cannot read them. The add-on does not run an Inference Bridge backend. Remote provider services may charge separately. Experimental decisions are available by default for separately approved typed questions over JSON state via cloud OpenRouter or local Ollama. Enable experimental speech in Options for separately approved transcription and MP3 speech synthesis. Firefox does not offer Chrome's On-device Prompt API.