LangChain

LangChain’s OpenAI integration works with TOKEN once the base URL points to https://api.thetoken.dev/v1.

Chat models

Python

pip install langchain-openai

import os
from langchain_openai import ChatOpenAI

llm = ChatOpenAI(
    model="zai-org/glm-5.3",
    base_url="https://api.thetoken.dev/v1",
    api_key=os.environ["TOKEN_API_KEY"],
    max_tokens=4096,
)

print(llm.invoke("Explain retries with backoff in two sentences.").content)

JavaScript

npm install @langchain/openai @langchain/core

import { ChatOpenAI } from '@langchain/openai'

const llm = new ChatOpenAI({
  model: 'zai-org/glm-5.3',
  apiKey: process.env.TOKEN_API_KEY,
  configuration: { baseURL: 'https://api.thetoken.dev/v1' },
  maxTokens: 4096,
})

const reply = await llm.invoke('Explain retries with backoff in two sentences.')
console.log(reply.content)

Streaming (llm.stream(...)), tool binding (llm.bind_tools(...) / llm.bindTools(...)) and structured output (with_structured_output / withStructuredOutput) work on models that support them; check the models page before choosing one. meta-llama/llama-4-maverick has no tool calling.

Embeddings

Python

import os
from langchain_openai import OpenAIEmbeddings

embeddings = OpenAIEmbeddings(
    model="baai/bge-m3",
    base_url="https://api.thetoken.dev/v1",
    api_key=os.environ["TOKEN_API_KEY"],
    check_embedding_ctx_length=False,
)

vector = embeddings.embed_query("Where is my order?")
print(len(vector))  # 1024

JavaScript

import { OpenAIEmbeddings } from '@langchain/openai'

const embeddings = new OpenAIEmbeddings({
  model: 'baai/bge-m3',
  apiKey: process.env.TOKEN_API_KEY,
  configuration: { baseURL: 'https://api.thetoken.dev/v1' },
})

const vector = await embeddings.embedQuery('Where is my order?')
console.log(vector.length) // 1024

In Python, check_embedding_ctx_length=False makes LangChain send the text itself. Without it, LangChain splits the text with OpenAI’s tokenizer and sends token numbers, which only OpenAI’s own embedding models understand.