Ranking list
OpenAI Compatible通常只表示请求格式接近，不代表所有Model、Tool Calling、流式输出和错误码完全一致。迁移前应做最小调用测试。
Suitable for developers wanting multi-model access via a single OpenAI-compatible endpoint. Prices, providers and available models change frequently. Verify model pages before integration.
Pricing Page把 Serverless token、GPU 时长和训练 token 分开Description，适合需要比较不同部署Method的用户。
Billing docs clearly state token prices, cache hit rules and balance deduction. Confirm current model names and top-up availability before integration.
适合需要 Qwen、DashScope 或阿里云账户体系的团队。费用项较多，接入前应分项核对。
适合Open-source Models、专属端点和Batch Inference场景。Model价格与限制需要在当前定价页和控制台共同核对。
适合需要 OpenAI CompatibleEndpoint和较多Open-source Models选择的开发者。
适合希望使用 DeepSeek、Qwen、GLM 等Model的中文开发者。Model价格和可用性应以当前页面为准。
覆盖 LLM、Image、视频等多种 API，Billing单位差异较大，适合逐项Verified后使用。
Provider docs show a unified AI model gateway, OpenAI SDK compatible endpoint, model pricing page and top-up instructions. Real-time prices, account limits and data processing boundaries need to be verified on the current page.
Provider pages show AI API and documentation, with references to OpenAI, Claude, Gemini etc. Costs, available models and account rules need to be verified on the current page.