VernLLM

VernLLM is a lightweight resilience layer for LLM clients that handles retries, timeouts, circuit breakers, caching, rate limiting, observability and cross provider fallbacks for your calls. Works with OpenAI, Anthropic, Bedrock, Gemini, and 30+ providers. Get schema-validated outputs, cache coalescing, tool calling, and streaming too. Framework agnostic and adapter-based, so you can swap providers easily. Built to save teams the sprints and incidents of maintaining LLM infrastructure.