Cerebras + LangChain ultra-fast inference

Ultra-low latency LLM inference (up to 2k+ tokens/sec) in LangChain agents, RAG pipelines, and multi-step chains using Cerebras CS-3 hardware.