A minimal and opinionated interface and helper functions for AI provider fetch implementations.
ai-fetch provides a streamlined interface and helper functions to implement various AI provider clients, such as openai-fetch and anthropic-fetch, using a consistent and minimal approach.
- You want a fast and small client that doesn't patch fetch
- Supports all environments with native fetch: Node 18+, browsers, Deno, Cloudflare Workers, etc
- Provides a consistent interface for different AI providers
- You only need chat, completions, embeddings, and moderations
- Your runtime doesn't have native fetch support
- Your app can't handle native ESM code
- You need endpoints other than chat, completions, embeddings, and moderations
- You aren't concerned with library size or fetch patching
npm install ai-fetchThis package requires node >= 18 or an environment with fetch support.
This package exports ESM. If your project uses CommonJS, consider switching to ESM or use the dynamic import() function.
A client that implements an AIChatClient:
import { type BaseAIClient, type AIChatClient, createApiInstance } from 'ai-fetch';
class YourAIClient implements BaseAIFetchClient, AIChatClient {
name = 'YourAIClient'
api: KyInstance
constructor() {
// Create a fetch api client for your provider using the helper function provided.
this.api = createApiInstance({
//... provide your ai provider opts here
})
}
/** Create a completion for a chat message. */
async createChatCompletion(
params: ChatParams<ChatModel>,
opts?: RequestOpts
): Promise<ChatResponse> {
// implementation
}
/** Create a chat completion and stream back partial progress. */
async streamChatCompletion(
params: ChatStreamParams<ChatModel>,
opts?: RequestOpts
): Promise<ChatStreamResponse> {
// streaming implementation
}
}The AI Fetch API follows OpenAI very closely, so their reference documentation can generally be used. Everything is strongly typed, so you will know if anything is different as soon as TypeScript parses your code.
// Generate a single chat completion
client.createChatCompletion(params: ChatParams<SomeModel>): Promise<ChatResponse>;
// Stream a single completion via a ReadableStream
client.streamChatCompletion(params: ChatStreamParams<SomeModel>): Promise<ChatStreamResponse>;
// Generate one or more completions
client.createCompletions(params: CompletionParams<SomeModel>): Promise<CompletionResponse>;
// Stream a single completion via a ReadableStream
client.streamCompletion(params: CompletionStreamParams<SomeModel>): Promise<CompletionStreamResponse>;
// Generate one or more embeddings
client.createEmbeddings(params: EmbeddingParams<SomeModel>): Promise<EmbeddingResponse>
// Checks for potentially harmful content
client.createModeration(params: ModerationParams<SomeModel>): Promise<ModerationResponse>The type definitions are available through TSServer, and can be found here: type definitions.
MIT © Dexa