mirror of
https://github.com/ILoveBingLu/CipherTalk.git
synced 2026-06-14 17:05:07 +08:00
ILoveBingLu
107 lines
2.9 KiB
TypeScript
107 lines
2.9 KiB
TypeScript
import { BaseAIProvider, type ChatOptions, type ChatWithToolsOptions } from './base'
|
|
|
|
/**
|
|
* 通义千问提供商元数据
|
|
*/
|
|
export const QwenMetadata = {
|
|
id: 'qwen',
|
|
name: 'qwen',
|
|
displayName: '通义千问',
|
|
description: '阿里云出品，稳定可靠',
|
|
models: [
|
|
'Qwen Plus',
|
|
'Qwen Flash',
|
|
'Qwen Turbo',
|
|
'QwQ Plus',
|
|
'QwQ Flash',
|
|
'QwQ Turbo',
|
|
'Qwen 3 Omni Flash',
|
|
'Qwen 3 Omni Turbo',
|
|
'Qwen 3 Omni Flash Turbo',
|
|
'Qwen 3 Omni Flash Turbo Flash',
|
|
'Qwen 3 Next 80B Thinking',
|
|
'Qwen 3 Next 80B Instruct',
|
|
'DeepSeek V3.2',
|
|
'Kimi k2 Thinking',
|
|
'GLM 4.7'
|
|
],
|
|
pricing: '¥0.008/1K tokens',
|
|
pricingDetail: {
|
|
input: 0.008, // 0.008元/1K tokens
|
|
output: 0.008
|
|
},
|
|
website: 'https://dashscope.aliyun.com/',
|
|
logo: './AI-logo/qwen-color.svg'
|
|
}
|
|
|
|
const MODEL_MAPPING: Record<string, string> = {
|
|
'Qwen Plus': 'qwen-plus',
|
|
'Qwen Flash': 'qwen-flash',
|
|
'Qwen Turbo': 'qwen-turbo',
|
|
'QwQ Plus': 'qwq-plus',
|
|
'QwQ Flash': 'qwq-flash',
|
|
'QwQ Turbo': 'qwq-turbo',
|
|
'Qwen 3 Omni Flash': 'qwen3-omni-flash',
|
|
'Qwen 3 Omni Turbo': 'qwen3-omni-turbo',
|
|
'Qwen 3 Omni Flash Turbo': 'qwen3-omni-flash-turbo',
|
|
'Qwen 3 Omni Flash Turbo Flash': 'qwen3-omni-flash-turbo-flash',
|
|
'Qwen 3 Next 80B Thinking': 'qwen3-next-80b-a3b-thinking',
|
|
'Qwen 3 Next 80B Instruct': 'qwen3-next-80b-a3b-instruct',
|
|
'DeepSeek V3.2': 'deepseek-v3.2',
|
|
'Kimi k2 Thinking': 'kimi-k2-thinking',
|
|
'GLM 4.7': 'glm-4.7'
|
|
}
|
|
|
|
/**
|
|
* 通义千问提供商
|
|
*/
|
|
export class QwenProvider extends BaseAIProvider {
|
|
name = QwenMetadata.name
|
|
displayName = QwenMetadata.displayName
|
|
models = QwenMetadata.models
|
|
pricing = QwenMetadata.pricingDetail
|
|
|
|
constructor(apiKey: string) {
|
|
super(apiKey, 'https://dashscope.aliyuncs.com/compatible-mode/v1')
|
|
}
|
|
|
|
/**
|
|
* 获取真实模型ID
|
|
*/
|
|
private getModelId(displayName: string): string {
|
|
return MODEL_MAPPING[displayName] || displayName
|
|
}
|
|
|
|
protected resolveModelId(displayName: string): string {
|
|
return this.getModelId(displayName)
|
|
}
|
|
|
|
protected getChatRequestExtraParams(options?: ChatOptions): Record<string, unknown> {
|
|
return {
|
|
enable_thinking: options?.enableThinking !== false
|
|
}
|
|
}
|
|
|
|
protected getToolRequestExtraParams(options: ChatWithToolsOptions): Record<string, unknown> {
|
|
return {
|
|
enable_thinking: options.enableThinking !== false
|
|
}
|
|
}
|
|
|
|
/**
|
|
* 重写 chat 方法以使用映射后的模型ID
|
|
*/
|
|
async chat(messages: any[], options?: any): Promise<string> {
|
|
const modelId = this.getModelId(options?.model || this.models[0])
|
|
return super.chat(messages, { ...options, model: modelId })
|
|
}
|
|
|
|
/**
|
|
* 重写 streamChat 方法以使用映射后的模型ID
|
|
*/
|
|
async streamChat(messages: any[], options: any, onChunk: (chunk: string) => void): Promise<void> {
|
|
const modelId = this.getModelId(options?.model || this.models[0])
|
|
return super.streamChat(messages, { ...options, model: modelId }, onChunk)
|
|
}
|
|
}
|