Watch an LLM answer token by token with per-token confidence. every-other-token streams a reply from OpenAI, Anthropic, Gemini, OpenRouter or a local OpenAI-compatible server (Ollama and others), shows each token's confidence and perplexity from logprobs, and can rewrite every other token live. It runs as a CLI, a terminal UI or a local web UI. The mock provider needs no API key.

WWW: https://github.com/Mattbusel/Every-Other-Token
