OpenAI API compatible LLM endpoints

Standardized API interfaces that mimic OpenAI's schema, allowing developers to swap proprietary models for self-hosted or alternative LLMs by changing a single base URL.

OpenAI API compatible LLM endpoints have become the de facto industry standard for deploying open-source models. By replicating OpenAI's exact request and response schemas (such as the /v1/chat/completions and /v1/embeddings paths), runtimes like vLLM, Ollama, and LM Studio eliminate vendor lock-in. Developers can transition from cloud-hosted GPT-4o instances to local, self-hosted models like Llama 3.3 or Qwen 2.5 without rewriting their core application logic. This standardization simplifies infrastructure migration, reduces API costs, and guarantees data privacy by keeping sensitive workloads on-premises.

https://github.com/vllm-project/vllm