Hey folks - sharing here in case it would help others too! we just open-sourced Viteval, a JS/TS framework for running LLM evals 🚀 It works just like Vitest, so you can run evals locally or in CI without changing how you write tests. We built it after getting frustrated with existing options: • Framework lock-in (we didn’t want to live inside LangChain just to run tests) • JS support always felt like an afterthought • Burning tokens every run because datasets keep regenerating 🤦 With Viteval you get: ✅ Powered by Vitest (a runner you already know) ✅ Write evals like unit tests, locally or in CI ✅ Save & reuse datasets → no more paying for repeats Hope this helps your workflow too! 🙌