One API for leading models
One OpenAI-compatible API for the models currently listed on the public catalog.
Browse allThe Unified Interface
For Every Model
Better pricing, higher availability, no subscription.
200T+
Monthly Tokens
10M+
Global Users
80+
Providers
500+
Models
deepseek/deepseek-v4
Higher availability
Route a model id across configured upstream channels, with failover when a channel is unavailable.
Learn more
Throughput
Latency
Price and performance
Control cost while keeping latency low with edge-friendly deployments close to your users.
Learn moreCustom data policy
Fine-grained controls so requests only go to providers and models you trust.
View docsFeatured models ›
From the public catalog snapshotted at the last site build
deepseek-ai/DeepSeek-V4-Flash
by openai
Interface
OpenAI-compatible
deepseek-ai/DeepSeek-V4-Pro
by openai
Interface
OpenAI-compatible
Pro/moonshotai/Kimi-K2.6
by openai
Interface
OpenAI-compatible
Featured agent apps ›
Use a custom base URL in agents such as OpenCode, Hermes, and OpenClaw.
OpenCode
AI-native coding workspace from idea to launch
Hermes Agent
An autonomous agent that grows with you
OpenClaw
Open-source agent framework for real automation
Get started in minutes
1
Create an account
Sign up to start. You can create an org for your team later.
2
Buy credits
Credits work across models and providers.
4/1$100
3/30$10
3
Get an API key
Create a key and call the OpenAI-compatible API.
ZBStream_API_KEY
sk-bwr-demo_a1b2c3d4e5f6g7h8Latest blog
View all →
ZBStream 产品介绍：一个 API，接入全部大模型
ZBStream 是面向开发者的 LLM 网关：一个 OpenAI 兼容 API Key，统一路由、故障转移、计费与用量追踪，接入 DeepSeek、通义、Kimi 等主流模型，无需订阅。
2026年8月24日
一文打通 OpenAI 与 Anthropic 双接口：ZBStream API 实战指南
ZBStream 同时提供 OpenAI 兼容与 Anthropic 兼容两套接口。本文用 Python、Node.js、Go 三种语言演示两种协议的调用方式、流式输出与选型建议。
2026年8月23日
Unified LLM API: One Key for Multiple Models
A unified LLM API lets you call many models with one OpenAI-compatible key and base URL. How ZBStream exposes Chat Completions, model ids, and billing in one place.
2026年8月19日
What is an LLM Gateway? (vs Router & Proxy)
An LLM gateway sits between your app and model providers, giving you one API for routing, failover, billing, and usage tracking. How it differs from a router, proxy, and inference provider.
2026年8月19日
图工程详解：它是什么、何时该用、何时不该用
图工程详解：图到底是什么、一个能立刻让你的 AI 变快的假边测试、那个能收回成本的单一模式（菱形）、这些方法在哪里悄悄失效、什么时候图是错误工具，以及如何用几分钟亲手构建一个真正的图。
2026年8月18日
Connect Hermes to ZBStream: Quick Install & API Configuration Guide
Step-by-step tutorial on installing Hermes and configuring it with your Zbstream API URL and Key. Get your AI assistant up and running in minutes.
2026年8月17日
FAQ
What is an LLM gateway?
An LLM gateway is a proxy layer between your app and model providers. It exposes one API for routing, failover, billing, and usage tracking across many models.
What is ZBStream?
ZBStream is an LLM gateway for developers. One API key and an OpenAI-compatible endpoint let you call models from multiple providers without rewriting your client.
How is ZBStream different from calling a model vendor directly?
Direct vendor APIs lock you to one provider. ZBStream unifies authentication, model IDs, routing, and credit billing so you can switch or fail over without changing application code.
Does ZBStream work with the OpenAI SDK?
Yes. Point the OpenAI SDK at the ZBStream base URL, set your ZBStream API key, and send Chat Completions or Messages requests as usual.