Aabao.AI is a unified, OpenAI-compatible LLM API gateway. It gives developers a single API key and endpoint that work across three AI coding terminals — Claude Code, OpenAI Codex, and Gemini CLI — so you copy one configuration block and start in under a minute.

Is Aabao.AI compatible with the OpenAI API?

Yes. Aabao.AI exposes an OpenAI-compatible endpoint at https://api.aabao.ai/v1. You can point the official OpenAI SDK or any OpenAI-compatible client at this base_url, and it also supports the OpenAI Responses API and the Anthropic Messages API for Claude clients.

Can I use one API key across Claude Code, Codex, and Gemini CLI?

Yes. A single Aabao.AI API key works for all three platforms. You set the same key once and switch tools without re-applying for separate credentials — one key, three terminals.

How do I configure Claude Code, Codex, or Gemini CLI with Aabao.AI?

Each CLI needs one config file: Codex uses ~/.codex/config.toml with base_url https://api.aabao.ai/v1, Claude Code uses ~/.claude/settings.json with ANTHROPIC_BASE_URL https://api.aabao.ai, and Gemini CLI uses ~/.gemini/settings.json plus a ~/.gemini/.env file. Paste the provided block, replace sk-xxx with your key, and run the tool.

Which AI models does Aabao.AI support?

Aabao.AI routes to leading models from all three major providers, including OpenAI GPT-5.5, Anthropic Claude (Opus 4.8, Sonnet 4.6, Haiku 4.5), and Google Gemini 3.1 Pro. You choose the model name in your CLI config or API request.

一个密钥，接入任何 AI 编程终端

在任何终端里，直连
全球 AI 大模型

兼容绝大部分官方协议，复制一段配置即可上手。

小白也能看懂的最简配置，复制粘贴就能跑。

查看配置 🛡️ 真伪检测站

99.9%

SLA

<2s

TTFT

10K+

DEVS

gemini_curl.sh

curl https://api.aabao.ai/v1beta/models/gemini-3.5-flash:generateContent \
  -H "x-goog-api-key: $API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "contents": [{
      "parts":[{"text": "你好"}]
    }]
  }'

openai_curl.sh

curl https://api.aabao.ai/v1/responses \
  -H "Authorization: Bearer $API_KEY" \
  -H "content-type: application/json" \
  -d '{
    "model": "gpt-5.6-sol",
    "max_output_tokens": 1024,
    "input": "你好"
  }'

claude_curl.sh

curl https://api.aabao.ai/v1/messages \
  -H "x-api-key: $API_KEY" \
  -H "content-type: application/json" \
  -d '{
    "model": "claude-opus-5",
    "max_tokens": 1024,
    "messages": [
      {"role": "user", "content": "你好"}
    ]
  }'

一个密钥通用

同一个 API Key，三个平台都能用。换工具不换配置，省去重复申请。

稳定直连

国内可直接访问，自动负载均衡与重试，告别频繁报错和超时。

按量计费

用多少付多少，无最低消费。注册即送试用额度，先试后用。

Toolbox

工具箱

除 API 接入外，我们也在陆续做一些小而实用的 AI 周边工具，开箱即用、持续上新。

Copy & Paste

三大平台 · 最简配置

选择你用的工具，复制下面的配置到对应文件即可。把 sk-xxx 换成自己的 Key 就能用。

OpenAI Codex

~/.codex/config.toml

Claude Code

~/.claude/settings.json

Gemini CLI

~/.gemini/settings.json

OpenAI Codex CLI TOML

官方配置文档

~/.codex/config.toml

# 默认使用 aabao.ai 作为模型来源
model_provider = "aabao"
model          = "gpt-5.6-sol"

# 模型来源：填入接口地址，密钥从 auth.json 读取
[model_providers.aabao]
name     = "aabao"
base_url = "https://api.aabao.ai/v1"
wire_api = "responses"

~/.codex/auth.json

{
  "OPENAI_API_KEY": "sk-xxx"
}

两个文件都建好后，把 auth.json 里的 sk-xxx 换成你的 Key（没有该文件就新建一个），运行 codex 即可。

Claude Code JSON

官方配置文档

~/.claude/settings.json

{
  "env": {
    "ANTHROPIC_BASE_URL": "https://api.aabao.ai",
    "ANTHROPIC_API_KEY":  "sk-xxx",
    "CLAUDE_CODE_DISABLE_EXPERIMENTAL_BETAS":  "1",
    "CLAUDE_CODE_EXPERIMENTAL_AGENT_TEAMS":  "1",
    "ANTHROPIC_DEFAULT_HAIKU_MODEL": "claude-haiku-4-5-20251001",
    "ANTHROPIC_DEFAULT_SONNET_MODEL":  "claude-sonnet-5",
    "ANTHROPIC_DEFAULT_OPUS_MODEL":  "claude-opus-5"
  }
}

修改 sk-xxx ，保存后运行 claude 即可。

Gemini CLI JSON

官方配置文档

~/.gemini/settings.json

{
  "model": {
    "name": "gemini-3.1-pro-preview"
  },
  "security": {
    "auth": {
      "selectedType": "gemini-api-key"
    }
  }
}

~/.gemini/.env

# 接口地址与密钥写在 .env 里（官方新版不放 settings.json）
GOOGLE_GEMINI_BASE_URL=https://api.aabao.ai
GEMINI_API_KEY=sk-xxx

新版 Gemini CLI 已改为嵌套格式，接口地址和密钥需写在 .env 文件里。两个文件都建好后运行 gemini 即可。