⚠️ Warning

Keep your API Key secure. Do not commit it to version control or hardcode it in frontend code.

export TOKENFAB_API_KEY="your-api-key"
# Windows PowerShell
$env:TOKENFAB_API_KEY="your-api-key"

pip install openai

from openai import OpenAI

client = OpenAI(api_key="your-api-key", base_url="https://api.tokenfab.cn/v1")

resp = client.chat.completions.create(model="glm-5.2", messages=[{"role":"user","content":"Introduce the TokenFab platform"}])
print(resp.choices[0].message.content)

pip install openai

from openai import OpenAI
client = OpenAI(api_key="your-api-key", base_url="https://api.tokenfab.cn/v1")

try:
    r = client.chat.completions.create(model="glm-5.2", messages=[{"role":"user","content":"Hello"}])
except Exception as e:
    print(f"Error: {e}")

💡 Tip

OpenAI-compatible features can be used directly; for TokenFab extensions such as Thinking Mode and Web Search, follow the corresponding sections.

⚠️ Tip

Anthropic protocol currently supports the following models: glm-5.2, qwen3.7-max, qwen3.7-plus, kimi-k3

import anthropic

client = anthropic.Anthropic(
    api_key="your-api-key",
    base_url="https://api.tokenfab.cn/anthropic",
)

message = client.messages.create(
    model="glm-5.2",
    max_tokens=1024,
    messages=[{"role": "user", "content": "hello"}],
    thinking={"type": "disabled"},
)
print(message.content[0].text)

with client.messages.stream(
    model="glm-5.2",
    max_tokens=1024,
    messages=[{"role": "user", "content": "Tell me a story"}],
) as stream:
    for text in stream.text_stream:
        print(text, end="", flush=True)

export ANTHROPIC_BASE_URL="https://api.tokenfab.cn/anthropic"
export ANTHROPIC_AUTH_TOKEN="your-api-key"
export ANTHROPIC_DEFAULT_MODEL="glm-5.2"
export ANTHROPIC_DEFAULT_HAIKU_MODEL="glm-5.2"
export ANTHROPIC_SMALL_FAST_MODEL="glm-5.2"
export CLAUDE_CODE_DEFAULT_HAIKU_MODEL="glm-5.2"
export CLAUDE_CODE_SUBAGENT_MODEL="glm-5.2"

$env:ANTHROPIC_BASE_URL="https://api.tokenfab.cn/anthropic"
$env:ANTHROPIC_AUTH_TOKEN="your-api-key"
$env:ANTHROPIC_DEFAULT_MODEL="glm-5.2"
$env:ANTHROPIC_DEFAULT_HAIKU_MODEL="glm-5.2"
$env:ANTHROPIC_SMALL_FAST_MODEL="glm-5.2"
$env:CLAUDE_CODE_DEFAULT_HAIKU_MODEL="glm-5.2"
$env:CLAUDE_CODE_SUBAGENT_MODEL="glm-5.2"

Tip

The "Migrate from Existing Installation" section above contains the configuration variables. Get the API Key from the console.

cd /path/to/my-project
claude

openclaw onboard --install-daemon

curl -fsSL https://openclaw.ai/install.sh | bash

iwr -useb https://openclaw.ai/install.ps1 | iex

openclaw dashboard

openclaw terminal

openclaw terminal --model glm-5.2

curl -fsSL https://raw.githubusercontent.com/NousResearch/hermes-agent/main/scripts/install.sh | bash

{
  "models": [
    {
      "id": "deepseek-v4-pro",
      "name": "DeepSeek V4 Pro",
      "vendor": "TokenFab",
      "url": "https://api.tokenfab.cn/v1/chat/completions",
      "apiKey": "${{API_KEY}}",
      "maxInputTokens": 128000,
      "maxOutputTokens": 8192,
      "supportsToolCall": true,
      "supportsImages": false,
      "relatedModels": {
        "lite": "deepseek-v4-flash",
        "reasoning": "deepseek-v4-pro"
      }
    },
    {
      "id": "deepseek-v4-flash",
      "name": "DeepSeek V4 Flash",
      "vendor": "TokenFab",
      "url": "https://api.tokenfab.cn/v1/chat/completions",
      "apiKey": "${{API_KEY}}",
      "maxInputTokens": 128000,
      "maxOutputTokens": 8192,
      "supportsToolCall": true,
      "supportsImages": false
    }
  ],
  "availableModels": [
    "deepseek-v4-pro",
    "deepseek-v4-flash"
  ]
}

💡 Tip

The url in the config uses the OpenAI-compatible Chat Completions endpoint. Make sure apiKey contains your actual TokenFab API Key and that the placeholder is removed.

🏆 Best Overall — glm-5.2

1M ultra-large context, flagship performance.

💻 Coding — deepseek-v4-pro

Code generation, debugging, review.

⚡ High Concurrency — qwen3.7-max

Millisecond response, suitable for simple tasks like customer service and classification.

⚠️ Warning

Exceeding the context limit may result in degraded response quality or errors.

from openai import OpenAI

client = OpenAI(api_key="your-api-key", base_url="https://api.tokenfab.cn/v1")

stream = client.chat.completions.create(model="glm-5.2", messages=[{"role":"user","content":"Tell me a story"}], stream=True)
for chunk in stream:
    if chunk.choices:
        if chunk.choices[0].delta.content:
            print(chunk.choices[0].delta.content, end="", flush=True)

data: {"id":"chatcmpl-xxx","object":"chat.completion.chunk","created":1784950000,"model":"glm-5.2","choices":[{"index":0,"delta":{"role":"assistant","content":"Today"},"finish_reason":null}]}
data: {"id":"chatcmpl-xxx","object":"chat.completion.chunk","created":1784950000,"model":"glm-5.2","choices":[{"index":0,"delta":{},"finish_reason":"stop"}]}
data: [DONE]

from openai import OpenAI

client = OpenAI(api_key="your-api-key", base_url="https://api.tokenfab.cn/v1")

response = client.chat.completions.create(
    model="deepseek-v4-pro",
    messages=[{"role": "user", "content": "What are the latest AI industry trends in 2026?"}],
    extra_body={"enable_thinking":True}
)

print(response.choices[0].message.reasoning)

Note

Thinking Mode is available on models that support it, such as deepseek-v4-pro, glm-5.2, and qwen3.7-max; check the model capability configuration. In Thinking Mode, temperature and other parameters do not take effect.

tools = [{"type":"function","function":{"name":"get_weather","description":"Check weather","parameters":{"type":"object","properties":{"city":{"type":"string"}},"required":["city"]}}}]
messages = [{"role":"user","content":"What is the weather like in Beijing?"}]
r1 = client.chat.completions.create(model="glm-5.2", messages=messages, tools=tools)
tc = r1.choices[0].message.tool_calls[0]
              result = "Beijing is sunny today, 25°C"
messages.append(r1.choices[0].message)
messages.append({"role":"tool","tool_call_id":tc.id,"content":result})
r2 = client.chat.completions.create(model="glm-5.2", messages=messages)
print(r2.choices[0].message.content)

messages = [{"role":"user","content":"What is the weather like in Beijing?"}]
r1 = client.chat.completions.create(model="glm-5.2", messages=messages)
messages.append({"role":"assistant","content":r1.choices[0].message.content})
messages.append({"role":"user","content":"What about Shanghai?"})
r2 = client.chat.completions.create(model="glm-5.2", messages=messages)

r = client.chat.completions.create(model="glm-5.2", messages=[{"role":"user","content":"List 3 fruits and their prices"}],
    response_format={"type":"json_schema","json_schema":{"name":"fruits","schema":{"type":"object","properties":{"fruits":{"type":"array","items":{"type":"object","properties":{"name":{"type":"string"},"price":{"type":"number"}},"required":["name","price"]}}},"required":["fruits"]}}})
print(r.choices[0].message.content)

import asyncio
from openai import AsyncOpenAI

client = AsyncOpenAI(api_key="your-api-key", base_url="https://api.tokenfab.cn/v1")

async def ask(q):
    r = await client.chat.completions.create(model="glm-5.2", messages=[{"role":"user","content":q}])
    print(r.choices[0].message.content)

async def main():
    tasks = [ask(q) for q in ["Summarize AI", "Translate Hello", "Recommend 3 books"]]
    await asyncio.gather(*tasks)

asyncio.run(main())

💡 Tip

Suitable for batch processing. Make sure not to exceed model rate limits.

from openai import OpenAI

client = OpenAI(api_key="your-api-key", base_url="https://api.tokenfab.cn/v1")

response = client.chat.completions.create(
    model="glm-5.2",
    messages=[{"role": "user", "content": "What are the latest trends in AI in 2026?"}],
    extra_body={"enable_web_search": True}
)

print(response.choices[0].message.content)

Note

Web Search increases response latency. Only enable it when real-time information is needed. Search consumes additional tokens, which count toward your total usage.

curl -X POST https://api.tokenfab.cn/v1/chat/completions \
  -H "Authorization: Bearer your-api-key" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "glm-5.2",
    "messages": [
      {"role": "system", "content": "You are a translation assistant"},
      {"role": "user", "content": "Translate the following English to Chinese: Hello World"}
    ],
    "temperature": 0.7,
    "max_tokens": 1024
  }'

{
  "id": "chatcmpl-abc123",
  "object": "chat.completion",
  "model": "glm-5.2",
  "choices": [{
    "index": 0,
    "message": {
      "role": "assistant",
      "content": "Hello World"
    },
    "finish_reason": "stop"
  }],
  "usage": {
    "prompt_tokens": 28,
    "completion_tokens": 5,
    "total_tokens": 33
  }
}

curl https://api.tokenfab.cn/v1/models \
  -H "Authorization: Bearer your-api-key"

{"object":"list","data":[
  {"id":"aicc-doubao-seedance-2-0","object":"model","owned_by":"tokenfab"},
  {"id":"deepseek-v4-flash","object":"model","owned_by":"tokenfab"},
  {"id":"deepseek-v4-pro","object":"model","owned_by":"tokenfab"},
  {"id":"glm-5.2","object":"model","owned_by":"tokenfab"},
  {"id":"glm-5.3","object":"model","owned_by":"tokenfab"},
  {"id":"happyhorse-1.1-i2v","object":"model","owned_by":"tokenfab"},
  {"id":"happyhorse-1.1-r2v","object":"model","owned_by":"tokenfab"},
  {"id":"happyhorse-1.1-t2v","object":"model","owned_by":"tokenfab"},
  {"id":"kimi-k2.6","object":"model","owned_by":"tokenfab"},
  {"id":"kimi-k2.7-code","object":"model","owned_by":"tokenfab"},
  {"id":"kimi-k3","object":"model","owned_by":"tokenfab"},
  {"id":"occupancy-hold","object":"model","owned_by":"tokenfab"},
  {"id":"qwen3.7-max","object":"model","owned_by":"tokenfab"},
  {"id":"qwen3.7-max-2026-06-08","object":"model","owned_by":"tokenfab"},
  {"id":"qwen3.7-plus","object":"model","owned_by":"tokenfab"},
  {"id":"viduq3-pro","object":"model","owned_by":"tokenfab"},
  {"id":"viduq3-turbo","object":"model","owned_by":"tokenfab"}
]}

Note

For higher concurrency needs, contact technical support to upgrade. Exceeding limits returns HTTP 429.

{"error":{"code":"invalid_api_key","message":"Authentication failed, please check your API Key"}}

from langchain_openai import ChatOpenAI
llm = ChatOpenAI(model="glm-5.2", api_key="your-api-key", base_url="https://api.tokenfab.cn/v1")
response = llm.invoke("Introduce deep learning")

💡 Tip

Frameworks compatible with the OpenAI SDK, such as LlamaIndex and Semantic Kernel, can all be directly integrated.

curl -X POST https://api.tokenfab.cn/v1/videos \
  -H "Authorization: Bearer your-api-key" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "happyhorse-1.1-t2v",
    "prompt": "A golden puppy running on the grass, sunny, camera following",
    "seconds": "5",
    "size": "1080P"
  }'

# Response Example
# {"id":"vid-abc123","status":"queued","created_at":"2026-07-01T12:00:00Z"}

# Query task status
curl https://api.tokenfab.cn/v1/videos/vid-abc123 \
  -H "Authorization: Bearer your-api-key"

# Response Example (processing)
# {"id":"vid-abc123","status":"processing","progress":45}

# Response Example (completed)
# {"id":"vid-abc123","status":"completed","url":"https://cdn.tokenfab.cn/...mp4"}

# Download video
curl -O https://api.tokenfab.cn/v1/videos/vid-abc123/content \
  -H "Authorization: Bearer your-api-key"

Note

Video generation tasks can take a long time. Creation is not idempotent - check for existing tasks first on network timeout, do not blindly retry POST.

curl -X POST https://api.tokenfab.cn/v1/videos \
  -H "Authorization: Bearer your-api-key" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "happyhorse-1.1-i2v",
    "prompt": "Camera slowly pushes in, a girl slightly turns her head and smiles",
    "input_reference": "https://example.com/photo.jpg",
    "seconds": "5",
    "size": "1080P"
  }'

curl -X POST https://api.tokenfab.cn/v1/videos \
  -H "Authorization: Bearer your-api-key" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "happyhorse-1.1-r2v",
    "prompt": "The person in [Image 1] walks down the street, keeping the outfit style consistent",
    "reference_images": [
      "https://example.com/ref1.jpg",
      "https://example.com/ref2.jpg"
    ],
    "seconds": "5",
    "size": "1080P"
  }'

curl -X POST https://api.tokenfab.cn/v1/videos \
  -H "Authorization: Bearer your-api-key" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "viduq3-turbo",
    "prompt": "City skyline transitioning from day to night, lights gradually turning on",
    "input_reference": "https://example.com/first_frame.jpg",
    "last_frame": "https://example.com/last_frame.jpg",
    "seconds": "5",
    "size": "1080P"
  }'