# One API — Unified LLM API Gateway (Docker)

> One API is a self-hosted LLM API gateway: unify OpenAI/Claude/Gemini/DeepSeek endpoints, manage keys, and deploy via Docker in minutes (33.7k★).

## Install

Save as a script file and run:

## Quick Use

```bash
docker run --name one-api -d --restart always -p 3000:3000 -e TZ=Asia/Shanghai -v /home/ubuntu/data/one-api:/data justsong/one-api
# Optional MySQL: add SQL_DSN env var (see README)
docker-compose up -d  # alternative compose deployment
```

## Intro

One API is a self-hosted LLM API gateway: unify OpenAI/Claude/Gemini/DeepSeek endpoints, manage keys, and deploy via Docker in minutes (33.7k★).

**Best for:** Developers who want one endpoint for multiple providers, with centralized key management and predictable routing for agents

**Works with:** Docker (recommended), optional MySQL, and downstream clients that can point to a single OpenAI-compatible base URL

**Setup time:** 10-30 minutes

## Main

- **Multi-provider gateway**: positioned as a unified API layer for many model providers, so agents can talk to “one base URL” instead of custom clients per vendor.
- **Docker-first deployment**: the README provides runnable `docker run` commands (SQLite and MySQL variants) and a Compose option.
- **Operational knobs**: documents env vars and multi-node notes (session secret, sync frequency, Redis) for production-ish usage.
- **Good for agent fleets**: centralizes key/quotas so you can rotate credentials without touching every agent or tool.

### FAQ

- **Do I need MySQL?**: No. The README shows an SQLite-based docker run; use MySQL when you need multi-node or stronger persistence.
- **How do I upgrade safely?**: The README suggests a watchtower-based update command; test on a staging instance first.
- **What is the first integration?**: Point one client at the gateway base URL and verify a single model call before migrating everything.

## Source & Thanks

> Source: https://github.com/songquanpeng/one-api
> License: MIT
> GitHub stars: 33663 · forks: 6429

---

<!-- ZH -->

## Quick Use

```bash
docker run --name one-api -d --restart always -p 3000:3000 -e TZ=Asia/Shanghai -v /home/ubuntu/data/one-api:/data justsong/one-api
# 可选 MySQL：按 README 加 SQL_DSN 环境变量
docker-compose up -d  # 也可用 compose 部署
```

## Intro

One API 是自建 LLM API 网关：统一 OpenAI/Claude/Gemini/DeepSeek 等接口，集中管理 key/额度，并可用 Docker 快速部署（33663★）。

**Best for:** 想把多个供应商收敛成一个统一入口，并集中管理 key/配额，让 agent 调用更可控的开发者

**Works with:** Docker（推荐）、可选 MySQL，以及能配置统一 OpenAI-compatible base URL 的下游客户端/应用

**Setup time:** 10-30 minutes

## Main

- **多供应商统一入口**：把多家模型提供方收敛成一个 API 层，让 agent 只需要配置一个 base URL。
- **Docker 优先**：README 给出可直接运行的 `docker run`（SQLite/MySQL）与 Compose 部署方式。
- **生产化参数**：文档包含环境变量、多机部署注意事项（SESSION_SECRET、同步、Redis 等）。
- **适合 agent fleet**：集中管理 key/额度，轮换凭据不用逐个改每个 agent/工具。

### FAQ

- **必须用 MySQL 吗？**: 不必。README 有 SQLite 的 docker run；需要多机/更强持久化时再上 MySQL。
- **怎么安全升级？**: README 给出 watchtower 更新命令；建议先在 staging 验证再升级生产。
- **先集成哪一步？**: 先让一个客户端指向 gateway base URL，跑通一次调用，再逐步迁移。

## Source & Thanks

> Source: https://github.com/songquanpeng/one-api
> License: MIT
> GitHub stars: 33663 · forks: 6429


---
Source: https://tokrepo.com/en/workflows/one-api-unified-llm-api-gateway-docker
Author: AI Open Source