Chat Worm

Chat Worm is a chat with images tool. Open source chat client that connects to OpenAI, Claude, Gemini, DeepSeek, Qwen, Grok, Mistral and StepFun through your own API keys, with model switching, voice conversation, DALL-E 3 images and vision analysis.

Chat Worm is a lightweight chat front end that connects to large language models through their official APIs rather than through a reseller. The premise is simple: you obtain an API key from a provider such as OpenAI, Anthropic, Google Gemini, DeepSeek, Alibaba Qwen, xAI Grok, Mistral AI or StepFun, paste it into the app, and start chatting. Because billing runs through the provider's API rather than a flat subscription, the project positions itself as a cheaper route to the same underlying models, and it avoids the message limits that come with consumer chat plans. The practical appeal is model switching. A single dropdown covers every provider you have connected, so when one API is slow or unavailable, another model can take over mid workflow. Settings expose the parameters that matter to power users: randomness (temperature) with a default of 1, a maximum token field, and a reasoning effort control with high, medium and low options for models that support it. Voice is handled through speech to text using Whisper and text to speech with six voice choices (Alloy, Echo, Fable, Onyx, Nova and Shimmer), and there is an option to send automatically after dictation so no click is needed. Image work is covered too, with DALL-E 3 generation and picture analysis through vision capable models. The audience is people who already understand API keys: developers, tinkerers and heavy chat users who want one interface across several vendors and do not mind managing their own credentials and spend. It will feel like extra work to anyone who wants a single monthly bill and a polished consumer assistant. Chat Worm is open source on GitHub and distributed as a web app, an Android app with over 1500 downloads, and a Windows app with over 500 downloads. Feature requests and bug reports go through the project's GitHub issue tracker, which is the natural support channel for a tool of this kind.