Tencent TokenHub

https://www.tencentcloud.com/products/tokenhub

我們支援所有 Tencent TokenHub 模型，只要在發送 completion 請求時將 tencent/ 設為前綴

TokenHub 是 Tencent Cloud 的統一 LLM 閘道。它提供與 OpenAI 相容的 Chat Completions endpoint，以及與 Anthropic 相容的 Messages endpoint，讓您能透過單一 API 金鑰存取 DeepSeek、GLM、Kimi、MiniMax 和 Hunyuan 模型。

API 金鑰

# env variable
os.environ['TENCENT_API_KEY']

範例用法

from litellm import completion
import os

os.environ['TENCENT_API_KEY'] = ""
response = completion(
    model="tencent/deepseek-v4-pro",
    messages=[
       {"role": "user", "content": "hello from litellm"}
   ],
)
print(response)

範例用法 - 串流

from litellm import completion
import os

os.environ['TENCENT_API_KEY'] = ""
response = completion(
    model="tencent/deepseek-v4-pro",
    messages=[
       {"role": "user", "content": "hello from litellm"}
   ],
    stream=True
)

for chunk in response:
    print(chunk)

支援的模型

我們支援 TokenHub international endpoint 上可用的所有模型。

模型名稱	函式呼叫
deepseek-v4-flash-202605	completion(model="tencent/deepseek-v4-flash-202605", messages)
deepseek-v4-pro-202606	completion(model="tencent/deepseek-v4-pro-202606", messages)
deepseek-v4-flash	completion(model="tencent/deepseek-v4-flash", messages)
deepseek-v4-pro	completion(model="tencent/deepseek-v4-pro", messages)
deepseek-v3.2	completion(model="tencent/deepseek-v3.2", messages)
glm-5.1	completion(model="tencent/glm-5.1", messages)
glm-5v-turbo	completion(model="tencent/glm-5v-turbo", messages)
glm-5-turbo	completion(model="tencent/glm-5-turbo", messages)
glm-5	completion(model="tencent/glm-5", messages)
kimi-k2.6	completion(model="tencent/kimi-k2.6", messages)
kimi-k2.5	completion(model="tencent/kimi-k2.5", messages)
minimax-m3	completion(model="tencent/minimax-m3", messages)
minimax-m2.7	completion(model="tencent/minimax-m2.7", messages)
minimax-m2.5	completion(model="tencent/minimax-m2.5", messages)
hy-mt2-plus	completion(model="tencent/hy-mt2-plus", messages)

自訂 API Base

預設情況下，LiteLLM 使用新加坡區域的 endpoint。您可以用 TENCENT_API_BASE 覆寫它。

import os

os.environ['TENCENT_API_BASE'] = "https://tokenhub.tencentcloudmaas.com/v1"  # Guangzhou region

Thinking / Reasoning 模式

許多 TokenHub 模型支援延伸思考。LiteLLM 支援 thinking 和 reasoning_effort 這兩個參數。

thinking 參數

from litellm import completion
import os

os.environ['TENCENT_API_KEY'] = ""

resp = completion(
    model="tencent/deepseek-v4-pro",
    messages=[{"role": "user", "content": "What is 2+2?"}],
    thinking={"type": "enabled"},
)

print(resp.choices[0].message.reasoning_content)
print(resp.choices[0].message.content)

reasoning_effort 參數

from litellm import completion
import os

os.environ['TENCENT_API_KEY'] = ""

resp = completion(
    model="tencent/deepseek-v4-pro",
    messages=[{"role": "user", "content": "What is 2+2?"}],
    reasoning_effort="medium",
)

print(resp.choices[0].message.reasoning_content)
print(resp.choices[0].message.content)

備註

當 reasoning_effort 的值不是 "none" 時，LiteLLM 會自動將其對應為 thinking={"type": "enabled"}。

基本用法

SDK

from litellm import completion
import os

os.environ['TENCENT_API_KEY'] = ""
resp = completion(
    model="tencent/deepseek-v4-pro",
    messages=[{"role": "user", "content": "Tell me a joke."}],
)

print(
    resp.choices[0].message.reasoning_content
)

PROXY

1. 設定 config.yaml

model_list:
  - model_name: deepseek-v4-pro
    litellm_params:
        model: tencent/deepseek-v4-pro
        api_key: os.environ/TENCENT_API_KEY

2. 執行 proxy

python litellm/proxy/main.py

3. 測試看看！

curl -L -X POST 'http://0.0.0.0:4000/v1/chat/completions' \
-H 'Content-Type: application/json' \
-H 'Authorization: Bearer sk-1234' \
-d '{
    "model": "deepseek-v4-pro",
    "messages": [
      {
        "role": "user",
        "content": "hello from litellm proxy"
      }
    ]
}'

與 Anthropic 相容的 Messages API

TokenHub 也提供與 Anthropic 相容的 Messages API。LiteLLM 會在可用時透過此 endpoint 路由請求。

os.environ['TENCENT_API_KEY'] = ""

若要分別覆寫與 Chat Completions endpoint 不同的 Anthropic 相容 base URL：

os.environ['TENCENT_ANTHROPIC_API_BASE'] = "https://tokenhub-intl.tencentcloudmaas.com"

當 TENCENT_ANTHROPIC_API_BASE 和 TENCENT_API_BASE 都有設定時，針對 Messages API 呼叫會以 Anthropic 專用的設定為優先。