// the find
sparckles/Robyn
Robyn is a Super Fast Async Python Web Framework with a Rust runtime.
Robyn is an async Python web framework where routing and request dispatch run through a Rust core, aiming at throughput closer to Node/Go while keeping a Flask-style decorator API. It's for people who want more raw speed than FastAPI/Starlette without leaving Python, and who can tolerate a framework that's still pre-1.0.
Supports both sync and async handlers on the same app without a separate adapter layer — the Rust runtime does the dispatch, which is a genuinely different mechanism than wrapping WSGI/ASGI. Multi-process and multi-worker scaling is a CLI flag (--processes, --workers) instead of something you bolt on with gunicorn afterward, which lowers the setup cost for squeezing one box before reaching for a load balancer. CI runs codspeed.yml for performance regression tracking on every change, not just a one-time benchmark screenshot in the README, so the speed claims actually get tested rather than just marketed. WebSockets, SSE, and streaming responses are first-party, so you're not pulling in a second library for anything beyond plain request/response.
Still pre-1.0 (0.6x releases) with no visible semver policy — expect breaking changes between minor versions, which matters for anything you plan to run for years. The feature list has grown to include an MCP server and 'AI Agent' routing bolted onto what is fundamentally a web framework; that reads as chasing whatever's trending rather than hardening the routing/middleware core that's the actual reason to pick this over FastAPI. Errors can originate in the Rust layer, and a Python traceback won't show you what happened there — debugging gets harder a step sooner than in a pure-Python framework. Building from source needs Rust, a C compiler, and maturin; fine on a wheel-supported platform, a real tax if you're not or need an unreleased fix.