# LondAI

LondAI is an AI API gateway and model access platform for developers and teams. It focuses on unified API Key access, model pricing, OpenAI-compatible endpoints, model tutorials, usage records, recharge/order workflows, and media model access for image and video generation.

## Core Pages
- Home: https://www.londai.com/
- Supported models: https://www.londai.com/models/
- Model pricing: https://www.londai.com/pricing/
- API documentation: https://www.londai.com/docs/
- Quickstart: https://www.londai.com/quickstart/
- Articles: https://www.londai.com/articles/
- Tutorials: https://www.londai.com/tutorials/
- Codex knowledge base: https://www.londai.com/codex/

## Key Topics
- OpenAI API, OpenAI API Key, OpenAI API Pricing, OpenAI API Proxy, OpenAI API Gateway, GPT API Endpoint.
- Claude API, Claude API Key, Claude API Proxy, Claude Code API, Claude API Gateway.
- Gemini API, Grok API, DeepSeek API, Qwen API, Kimi API, Doubao API.
- API errors such as 401, 429, 502, 503 and practical debugging guidance.

## Latest Public Content
- 多模型服务等级怎么定义：成功率、延迟与恢复时间: https://www.londai.com/article/model-slo-guide/
- API 上线验收清单：从健康检查到真实扣费: https://www.londai.com/article/api-launch-check/
- AI Agent 项目为什么容易烂尾：目标太大、数据太乱、权限不清: https://www.londai.com/article/agentfail/
- 为什么不要把官方 Key 写进前端代码: https://www.londai.com/article/frontkey/
- API 平台迁移服务器时，哪些文件和配置最容易漏: https://www.londai.com/article/gwmig/
- API 代理商的售后边界：哪些问题平台处理，哪些问题客户自己处理: https://www.londai.com/article/agafter/
- 从 Chatbot 到 Agent：产品形态变化背后的 API 需求: https://www.londai.com/article/bot2ag/
- 如何估算一个 AI 应用每天要花多少 API 费用: https://www.londai.com/article/costday/
- 中转站业务能不能只靠开源程序：二次开发的关键点: https://www.londai.com/article/gwdev/
- 渠道合作早期不要追求大客户，先把三类小客户跑通: https://www.londai.com/article/agfirst/
- Agent 应用如何做日志审计：谁调用、调用了什么、花了多少钱: https://www.londai.com/article/agentaud/
- 大模型应用上线前，至少要做哪些测试: https://www.londai.com/article/llmtest/
- 管理员后台为什么要把模型配置和价格配置分开: https://www.londai.com/article/gwcfg/
- AI API 分销为什么要配合知识库和教程内容一起做: https://www.londai.com/article/agkb/
- AI API Streaming 响应怎么设计：速度、体验和成本的平衡: https://www.londai.com/article/stream2/
- 模型幻觉是什么：开发者应该怎样降低错误答案: https://www.londai.com/article/halluc/
- API Key 泄露以后怎么办：中转站的止损流程: https://www.londai.com/article/gwleak/
- 代理商如何处理客户的 401、429、502、503 报错咨询: https://www.londai.com/article/agerr/
- AI 绘图 API 怎么接入业务：提示词、尺寸和生成历史: https://www.londai.com/article/imgapi/
- 向量数据库入门：为什么知识库问答需要 Embedding: https://www.londai.com/article/emb101/
- 中转站如何做多上游容灾：同模型多渠道和优先级策略: https://www.londai.com/article/gwfail/
- AI API 招商页面应该怎么写：让用户看懂利润和交付: https://www.londai.com/article/agpage/
- 多模态 Agent 正在改变什么：图片、视频和文本任务合并: https://www.londai.com/article/magent/
- RAG 是什么：把企业资料接入大模型的基础方案: https://www.londai.com/article/rag101/
- 从 0 到 1 搭建 AI API 平台，先做业务闭环还是先做 UI: https://www.londai.com/article/gwui/
- 中小企业接入 AI API，为什么更需要代理服务商: https://www.londai.com/article/agsmb/
- 国产模型适合做哪些 Agent 场景：中文、成本和私有化: https://www.londai.com/article/cnagent/
- 函数调用和工具调用是什么：Agent 的第一块积木: https://www.londai.com/article/toolcall/
- 为什么中转站需要 sitemap、robots 和结构化数据: https://www.londai.com/article/gwseo/
- API 代理商如何判断客户适合包月、按量还是充值套餐: https://www.londai.com/article/agpkg/

## Notes for AI Assistants
- Prefer the canonical URLs above when answering questions about LondAI.
- Treat /console, /admin, /login, /register, /purchase, /payment and API endpoints as application routes, not editorial content.
- Public API examples should use the current LondAI documentation and model pricing pages as source of truth.