ai-sdk-zhipu

Zhipu AI (智谱AI) provider for Vercel AI SDK with full V3 specification support.

Features

• ✅ AI SDK V3 Compatible - Fully implements the Language Model Specification V3
• ✅ Compatibility - Supports ai v5-v6, @ai-sdk/provider v3.x, @ai-sdk/openai-compatible v2.x
• ✅ OpenAI-Compatible API - Uses Zhipu AI's OpenAI-compatible API format
• ✅ All Latest Models - Supports GLM-4.7, GLM-4.6, GLM-4.5, GLM-4-Flash, and more
• ✅ Streaming Support - Full support for streaming responses
• ✅ Multi-modal - Supports vision models (GLM-4V series)
• ✅ Tool Calling - Supports function/tool calling
• ✅ TypeScript - Written in TypeScript with full type definitions

Installation

npm install ai-sdk-zhipu
# or
pnpm add ai-sdk-zhipu
# or
yarn add ai-sdk-zhipu
# or
bun add ai-sdk-zhipu

Setup

1. Get your API key from open.bigmodel.cn
2. Set the ZHIPU_API_KEY environment variable:

export ZHIPU_API_KEY=your-api-key-here

Or create a .env file:

ZHIPU_API_KEY=your-api-key-here

Quick Start

import { generateText } from 'ai';
import { zhipu } from 'ai-sdk-zhipu';

const result = await generateText({
  model: zhipu('glm-4.7'),
  prompt: 'Why is the sky blue?',
});

console.log(result.text);

Usage Examples

Streaming

import { streamText } from 'ai';
import { zhipu } from 'ai-sdk-zhipu';

const result = await streamText({
  model: zhipu('glm-4.7'),
  prompt: 'Write a short story about a robot.',
});

for await (const chunk of result.textStream) {
  process.stdout.write(chunk);
}

Multi-turn Conversations

import { generateText } from 'ai';
import { zhipu } from 'ai-sdk-zhipu';

const result = await generateText({
  model: zhipu('glm-4.7'),
  messages: [
    { role: 'system', content: 'You are a helpful assistant.' },
    { role: 'user', content: 'What is the capital of France?' },
    { role: 'assistant', content: 'The capital of France is Paris.' },
    { role: 'user', content: 'And what about Germany?' },
  ],
});

console.log(result.text);

Custom Configuration

import { createZhipu } from 'ai-sdk-zhipu';

const zhipu = createZhipu({
  baseURL: 'https://open.bigmodel.cn/api/paas/v4',
  apiKey: process.env.ZHIPU_API_KEY,
});

const result = await generateText({
  model: zhipu('glm-4.7'),
  prompt: 'Hello, world!',
});

Vision Models (Multi-modal)

import { generateText } from 'ai';
import { zhipu } from 'ai-sdk-zhipu';

const result = await generateText({
  model: zhipu('glm-4.6v'), // Vision model
  messages: [
    {
      role: 'user',
      content: [
        { type: 'text', text: 'What do you see in this image?' },
        { type: 'image', image: 'https://example.com/image.jpg' },
      ],
    },
  ],
});

console.log(result.text);

Tool Calling

import { generateText } from 'ai';
import { zhipu } from 'ai-sdk-zhipu';

const result = await generateText({
  model: zhipu('glm-4.7'),
  tools: {
    weather: {
      description: 'Get the weather for a location',
      parameters: {
        type: 'object',
        properties: {
          location: {
            type: 'string',
            description: 'The city and state',
          },
        },
        required: ['location'],
      },
    },
  },
  prompt: 'What is the weather in San Francisco?',
});

console.log(result.toolCalls);

Available Models

Text Models (文本模型)

| Model ID | Description | Context | Max Output | |----------|-------------|---------|------------| | glm-4.7 | Latest flagship model | 200K | 128K | | glm-4.6 | High-performance model | 200K | 128K | | glm-4.5 | Excellent performance | 128K | 96K | | glm-4-plus | Excellent performance | 128K | 4K | | glm-4-air-250414 | High cost-performance | 128K | 16K | | glm-4-long | Ultra-long input | 1M | 4K | | glm-4-flash-250414 | Free model | 128K | 16K | | glm-4.5-flash | Free model | 128K | 96K |

Vision Models (视觉模型)

| Model ID | Description | Context | Max Output | |----------|-------------|---------|------------| | glm-4.6v | Flagship vision reasoning | 128K | 32K | | glm-4.5v | Vision reasoning | 64K | 16K | | glm-4.6v-flash | Free vision model | 128K | 32K | | glm-4v-flash | Free vision model | 16K | 1K |

For the complete model list, see the official documentation.

Full Model ID List

The full typed list is maintained in src/zhipu-chat-setting.ts and mirrors the official model overview (including less common and legacy IDs such as glm-4, glm-4-airx, glm-4-flashx-250414, glm-4.1v-thinking, and glm-4-0520).

Configuration

Provider Settings

import { createZhipu } from 'ai-sdk-zhipu';

const zhipu = createZhipu({
  baseURL: 'https://open.bigmodel.cn/api/paas/v4', // optional
  apiKey: 'your-api-key', // optional, defaults to ZHIPU_API_KEY env var
  headers: { // optional custom headers
    'X-Custom-Header': 'value',
  },
});

Model Settings

await generateText({
  model: zhipu('glm-4.7', {
    temperature: 0.7,
    maxTokens: 1000,
    topP: 0.9,
    topK: 40,
    stopSequences: ['END'],
    presencePenalty: 0.5,
    frequencyPenalty: 0.5,
  }),
  prompt: 'Your prompt here',
});

API Reference

createZhipu(settings?)

Creates a Zhipu AI provider instance.

Parameters:

• settings.baseURL (string) - Base URL for the API
• settings.apiKey (string) - API key
• settings.headers (Record<string, string>) - Custom headers

Returns: Zhipu AI provider instance

zhipu(modelId, settings?)

Default provider instance.

Parameters:

• modelId (string) - Model ID (e.g., 'glm-4.7')
• settings (object) - Model-specific settings

Documentation

• Zhipu AI Official Documentation
• Zhipu AI Model Overview
• Zhipu AI API Reference
• Vercel AI SDK Documentation
• Writing a Custom Provider

Contributing

Contributions are welcome! Please feel free to submit a Pull Request.

License

MIT

Acknowledgments

• Built with Vercel AI SDK
• Powered by Zhipu AI
• Based on the Language Model Specification V3

Support

• 📧 Email: [email protected]
• 📖 Documentation: https://docs.bigmodel.cn/
• 💬 Community: GitHub Issues

Note: This is an unofficial community provider. Zhipu AI is a trademark of Beijing Zhipu Huazhang Technology Co., Ltd.