OpenAI 兼容接口 · 学 AI

解决什么问题What problem it solves

本地跑的模型，用现成的 OpenAI 客户端就能连。Serve a local model so existing OpenAI clients can connect unchanged.

这是什么What it is

vLLM 官方提供兼容 OpenAI Completions、Chat Completions、Embeddings 等接口的 HTTP 服务。

llama.cpp 的服务端同样以 OpenAI 兼容 API 方式启动。

vLLM provides an HTTP server compatible with OpenAI's Completions, Chat Completions and Embeddings APIs.

llama.cpp's server likewise launches as an OpenAI-compatible API server.

为什么需要知道Why it matters

换后端不用改代码，这是本地与云端之间最省事的切换方式。Swap backends without touching your code: the cheapest switch between local and cloud.

别名Chinese aliases

OpenAI 兼容 API

接着读什么，为什么What to read next, and why

吞吐Throughput, TPS
一起用于《接口兼容就等于行为一致》Used together in “A Compatible API Means Identical Behaviour”
迁移评估Migration Evaluation
一起用于《换模型前怎么迁移评估》Used together in “How to Assess a Model Migration”
模型版本与变体Model Family, Version, Variant
一起用于《换模型前怎么迁移评估》Used together in “How to Assess a Model Migration”

现在能做什么What you can do now

出处Sources

OpenAI 兼容接口OpenAI-compatible API