deepreasoning

Why we included this project

Instead of juggling two providers, DeepReasoning runs as a self-hosted Rust service that returns a single streaming response: DeepSeek R1 emits its chain-of-thought reasoning trace first, then Claude picks up and writes the final answer. You get one OpenAI-style API endpoint and a bundled chat web UI, so your application never has to stitch the two model outputs together itself. That arrangement suits teams that want R1's visible reasoning on their own infrastructure without maintaining separate integrations for each model. Because it keeps API keys and data local and exposes a familiar chat-completions interface, it drops into existing pipelines or frontends with minimal glue code.