This skill provides complete API calling information for mainstream domestic and international large models, helping you quickly find how to call any large model.
Quick Start
This skill is triggered when users ask the following questions:
- "How do I call the OpenAI/GPT API?"
- "What is the request format for the Claude API?"
- "What multimodal features does Gemini support?"
- "Which domestic model supports video generation?"
- "How do I use Doubao's Seedream?"
Supported Models
International Models
- OpenAI - GPT-5.2, GPT-4.1, GPT-Image, Sora
- Anthropic - Claude 4.5, Claude 4 series
- Google - Gemini 3, Gemini 2.5
Domestic Models
- Alibaba Cloud Model Studio - Tongyi Qianwen series
- Baidu Qianfan - ERNIE, DeepSeek, etc.
- ByteDance Doubao - doubao series + Seedream/Seedance
- Tencent Hunyuan - hunyuan series
- Zhipu GLM - GLM-4.7 series
- MiniMax - M2.1 series
- Moonshot AI - Kimi K2 series
Contents
Refer to the corresponding reference documentation based on the model the user asks about:
- OpenAI - GPT series, image, video, and speech APIs
- Anthropic - Claude series APIs
- Google - Gemini series APIs
- Alibaba Cloud Model Studio - Tongyi Qianwen series APIs (including multimodal)
- Baidu Qianfan - ERNIE Bot, ERNIE series (including Steam audio/video)
- ByteDance Doubao - Doubao series, Seedream image generation, Seedance video generation
- Tencent Hunyuan - Hunyuan series, image generation
- Zhipu GLM - GLM-4.7 series (including thinking mode)
- MiniMax - M2.1 series, speech/video/music generation
- Moonshot AI Kimi - K2 series, ultra-long context
General Information
Multimodal Capability Comparison
| Provider | Text | Image Input | Image Generation | Video Input | Video Generation | Speech Input | Speech Output |
|---|---|---|---|---|---|---|---|
| OpenAI | ✅ | ✅ | ✅ | ❌ | ✅ | ✅ | ✅ |
| Anthropic | ✅ | ✅ | ❌ | ❌ | ❌ | ❌ | ❌ |
| ✅ | ✅ | ✅ | ✅ | ✅ | ✅ | ✅ | |
| Alibaba Cloud | ✅ | ✅ | ✅ | ❌ | ❌ | ❌ | ❌ |
| Baidu | ✅ | ✅ | ✅ | ✅ | ✅ | ✅ | ✅ |
| ByteDance | ✅ | ✅ | ✅ | ✅ | ✅ | ✅ | ✅ |
| Tencent | ✅ | ✅ | ✅ | ❌ | ❌ | ❌ | ❌ |
| Zhipu | ✅ | ✅ | ✅ | ✅ | ✅ | ❌ | ❌ |
| MiniMax | ✅ | ❌ | ✅ | ❌ | ✅ | ✅ | ✅ |
| Kimi | ✅ | ✅ | ❌ | ❌ | ❌ | ❌ | ❌ |
OpenAI SDK-Compatible Interfaces
All domestic models support OpenAI SDK-compatible interfaces; simply modify base_url:
from openai import OpenAI
# 阿里云
client = OpenAI(
api_key="DASHSCOPE_API_KEY",
base_url="https://dashscope.aliyuncs.com/compatible-mode/v1"
)
# 百度
client = OpenAI(
api_key="QIANFAN_ACCESS_TOKEN",
base_url="https://qianfan.baidubce.com/v2"
)
# 字节
client = OpenAI(
api_key="ARK_API_KEY",
base_url="https://ark.cn-beijing.volces.com/api/v3"
)
# 腾讯
client = OpenAI(
api_key="HUNYUAN_API_KEY",
base_url="https://api.hunyuan.cloud.tencent.com/v1"
)
# 智谱
client = OpenAI(
api_key="ZHIPU_API_KEY",
base_url="https://api.z.ai/api/paas/v4"
)
# MiniMax
client = OpenAI(
api_key="MINIMAX_API_KEY",
base_url="https://api.minimaxi.com/v1"
)
# Kimi
client = OpenAI(
api_key="MOONSHOT_API_KEY",
base_url="https://api.moonshot.cn/v1"
)
Notes
- API Key Security: Do not expose API Keys in client-side code; use environment variables
- Region Selection: Choose the appropriate API endpoint based on the user's location (Beijing/Singapore/US)
- Free Quota: Most platforms offer free trial quota
- Real-Name Verification: Domestic API Keys usually require real-name verification
- Model Updates: Model versions are updated periodically. Information date: February 5, 2026
Disclaimer
The API information in this skill comes from each vendor's official public documentation and is for reference only. API specifications may change at any time; please refer to the latest official documentation.