Prompt Mixer

Prompt Mixer is an open-source desktop application for building, testing, and comparing prompts across multiple large language model providers. Users can chain prompts together so the output of one step feeds into the next, then evaluate results using built-in metrics such as regex matching, semantic similarity, or JSON schema validation, or write custom scoring scripts in JavaScript or Python. Datasets can be attached to test a prompt or chain against many inputs at once, and changes are tracked with version control. The tool connects to a wide range of model providers, including OpenAI, Anthropic, Google Gemini, DeepSeek, Perplexity, Cohere, AI21, Amazon Bedrock, Azure OpenAI, OpenRouter, and Ollama for running local models, plus DeepInfra, along with image-generation connectors for DALL-E and Stable Diffusion. It is built with Electron and TypeScript and runs on macOS, Windows, and Linux, targeting teams of engineers, product managers, and data specialists who need to collaborate on prompt development rather than working in isolated notebooks or chat windows. Prompt Mixer's core desktop app is free and licensed under the MIT license, with the source code hosted on GitHub. The project is still pre-1.0, with a relatively small open-source community, so some features and provider connectors may be less mature than in longer-established commercial tools.