Tianshu Docs

Outstanding value

Pay-as-you-go with transparent pricing and consistently low rates. Pick from multiple tiers and spend every cent where it matters.

A vast model lineup

Aggregates mainstream models such as Claude, GPT, Codex and image generation, with new ones added continuously — one key calls them all.

Standard interface

Fully OpenAI-compatible. Most clients, CLIs and SDKs work by simply changing the base URL — zero migration cost.

Stable and reliable

Multi-channel aggregation, smart scheduling and automatic failover, with a dedicated direct line as a fallback for long requests.

What is Tianshu

Tianshu (tian-shu.org) is a cost-effective LLM API aggregation platform. We bring many model providers together behind one unified standard interface, so all you need to do is:

Point your client / SDK base_url to Tianshu;

Call a vast range of large models just like you call OpenAI.

There's no need to integrate, top up, or maintain keys separately for each provider. For available models and live rates, see the Model Marketplace.

What can I use it for

Command-line coding: Connect through Claude Code or Codex for low-cost AI coding.

Graphical clients: Cherry Studio, Cline / Roo Code, Chatbox and any other OpenAI-compatible client.

AI image generation: image models such as GPT Image and Midjourney.

Your own programs: use the official OpenAI SDK and point base_url to Tianshu.

Three steps to get started

Step	What to do	Where
1	Sign up and add credit	Onboarding Steps
2	Create an API key (pick the right group)	Create a Key
3	Configure your client / CLI	Configure Clients

Step

What to do

Where

Onboarding Steps

Create an API key (pick the right group)

Create a Key

Configure your client / CLI

Configure Clients

If you're new, read Quick Start first; if you'd rather configure everything with a ready-made tool, jump straight to CC-Switch.

Tip

This site does not list specific prices — models and market rates change fast, so the live rates on the Model Marketplace are always authoritative.

Tianshu