Other clients & frameworks

Setup notes for common tools such as Cursor, Cline / Roo Code, LangChain, Dify and LobeChat.

Any tool that supports a "custom OpenAI endpoint" works. Remember two rules:

A few tools (such as Cherry Studio) append /v1 automatically; for those, enter only the root URL. If unsure, try both: a 404 usually means /v1 is extra or missing.

Cursor

Settings → Models:

1. Turn on OpenAI API Key and enter your key.
2. Turn on Override OpenAI Base URL and enter https://api.onefast.ai/v1.
3. Click "Add model" to add model names (e.g. gpt-5.5, claude-sonnet-4-6), save, then click Verify.

Some Cursor features (such as Tab completion) only use Cursor's own models and don't go through a custom endpoint; chat and Agent can use it.

Cline / Roo Code / Kilo Code (VS Code extensions)

• Set API Provider to OpenAI Compatible: Base URL https://api.onefast.ai/v1, API Key your key, Model ID the model name.
• Or choose Anthropic: enable the custom base URL and enter https://api.onefast.ai, for Claude models (supports prompt caching, which saves money).

LangChain

Python

# pip install langchain-openai
from langchain_openai import ChatOpenAI

llm = ChatOpenAI(model="gpt-5.5", api_key="sk-xxx", base_url="https://api.onefast.ai/v1")
print(llm.invoke("Hello!").content)

Python (Claude)

# pip install langchain-anthropic
from langchain_anthropic import ChatAnthropic

llm = ChatAnthropic(model="claude-sonnet-4-6", api_key="sk-xxx", base_url="https://api.onefast.ai")
print(llm.invoke("Hello!").content)

Dify / FastGPT / LobeChat / Open WebUI

Under "Model Provider", choose OpenAI-API-compatible (or "Custom OpenAI"):

• API Endpoint: https://api.onefast.ai/v1
• API Key: your key
• Model name: as listed in the model catalog

Plain HTTP

Any language can send HTTP requests directly. See the cURL example for each endpoint in the API reference, or download the OpenAPI file and import it into Postman / Apifox to generate code.