ai-fallback

AI Fallback

Automatically switch between AI model providers when one experiences downtime or errors.

[!warning] Version compatibility:

• Use ai-fallback version 0 for AI SDK v4.
  
• Use ai-fallback version 1 for AI SDK v5 (currently in beta).

Why?

AI providers can encounter:

• Rate limiting
• Service outages
• API errors
• Capacity issues
• Timeouts

This package ensures reliability by specifying multiple AI models as fallbacks. It automatically switches to the next available model if the primary fails, maintaining application uptime.

Installation

npm install ai-fallback@0 # use version 0.x.x for AI sdk v4

Models Fallback Reset

Reset to the primary model after a delay (e.g., 1 minute):

const model = createFallback({
    models: [
        anthropic('claude-3-haiku-20240307'), // Use Claude as preferred model
        openai('gpt-3.5-turbo'),
    ],
    onError: (error, modelId) => {
        console.error(`Error with model ${modelId}:`, error)
    },
    modelResetInterval: 60000, // Reset to first model after 1 minute of the first error
})

Usage

Create a Fallback Model

import { createFallback } from 'ai-fallback'
import { openai } from '@ai-sdk/openai'
import { anthropic } from '@ai-sdk/anthropic'

const model = createFallback({
    models: [anthropic('claude-3-haiku-20240307'), openai('gpt-3.5-turbo')],
})

Retry After Output

The retryAfterOutput option allows retrying with a different model even if some tokens were already streamed. This is useful when you want to restart the generation from scratch if an error occurs mid-stream:

import { createFallback } from 'ai-fallback'
import { openai } from '@ai-sdk/openai'
import { anthropic } from '@ai-sdk/anthropic'
import { streamText } from 'ai'

let fullText = ''

const model = createFallback({
    models: [anthropic('claude-3-haiku-20240307'), openai('gpt-3.5-turbo')],
    retryAfterOutput: true, // Enable retrying even after partial output
    onError: (err) => {
        console.error('Error:', err)
        // reset the full text because error happened when some tokens were already streamed in
        fullText = ''
    },
})

const stream = await streamText({
    model,
    system: 'You are a helpful assistant.',
    messages: [{ role: 'user', content: 'Write a long story.' }],
})

for await (const chunk of stream.textStream) {
    fullText += chunk
    console.log('Current text:', fullText)
}

Text Generation

Generate text with automatic fallback:

const result = await generateText({
    model,
    system: 'You are a helpful assistant.',
    messages: [{ role: 'user', content: 'Count from 1 to 5.' }],
})

Streaming Responses

Stream text responses:

const stream = await streamText({
    model,
    system: 'You are a helpful assistant.',
    messages: [{ role: 'user', content: 'Count from 1 to 5.' }],
})

for await (const chunk of stream.textStream) {
    console.log(chunk)
}

Structured Output

Stream typed objects using Zod schemas:

import { z } from 'zod'

const stream = await streamObject({
    model,
    system: 'You are a helpful assistant.',
    messages: [
        {
            role: 'user',
            content: 'Give me a person object with name and age properties.',
        },
    ],
    schema: z.object({
        name: z.string(),
        age: z.number(),
    }),
})

for await (const chunk of stream.partialObjectStream) {
    console.log(chunk)
}

Changelog

1.0.2

Patch Changes

• Handle anthropic errors after stream start as part errors