Generating

Generate text on Android with Gemini Nano: pass adk() to the AI SDK’s generateText or streamText. The examples on this page run on device. For cloud Gemini, create the model as shown in Models & Availability; everything else stays the same.

Text Generation

import { adk } from '@react-native-ai/adk'
import { generateText } from 'ai'

const { text } = await generateText({
  model: adk(),
  prompt: 'Explain quantum computing in simple terms',
})

The first call prepares Gemini Nano. To avoid the wait, call adk.prepareNano() ahead of time, once the availability check passes.

Streaming

import { adk } from '@react-native-ai/adk'
import { streamText } from 'ai'

const { textStream } = streamText({
  model: adk(),
  prompt: 'Write a short story about a robot learning to paint',
})

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

Streaming needs the polyfills.

Token Usage

usage reports Gemini’s token counts, on generateText results and in the finish part of a stream:

const { usage } = await generateText({
  model: adk(),
  prompt: 'Count to five.',
})

console.log(usage.inputTokens)
console.log(usage.outputTokens)
console.log(usage.raw?.totalTokenCount)

Tool Calling

Let the model call your JavaScript functions, on device or in the cloud. ADK runs the tool loop natively, so register each tool with the provider in availableTools, and pass the ones a request may use in tools as usual:

import { createAdkProvider } from '@react-native-ai/adk'
import { generateText, tool } from 'ai'
import { z } from 'zod'

const getCurrentTime = tool({
  description: 'Get the current time for a city',
  inputSchema: z.object({ city: z.string() }),
  execute: async ({ city }) => ({
    city,
    time: new Date().toLocaleTimeString(),
  }),
})

const adk = createAdkProvider({
  availableTools: { getCurrentTime },
})

const { text } = await generateText({
  model: adk(),
  tools: { getCurrentTime },
  prompt: 'What time is it in Warsaw?',
})

Because ADK runs the loop:

AI SDK step controls, such as stopWhen, don’t limit it.
Streamed tool calls are marked providerExecuted: true.
A required toolChoice is ignored with a warning.

Image Input

Ask about an image by passing it as a file part, as base64, a data URL or a Uint8Array:

const { text } = await generateText({
  model: adk(),
  messages: [
    {
      role: 'user',
      content: [
        { type: 'text', text: 'What is in this image?' },
        { type: 'file', mediaType: 'image/jpeg', data: base64Image },
      ],
    },
  ],
})

File and HTTP URLs aren’t supported yet, so read the image first.

JSON Output

Ask for JSON with Output.json():

import { generateText, Output } from 'ai'

const { output } = await generateText({
  model: adk(),
  prompt: 'Return a JSON object with name and age fields.',
  output: Output.json(),
})

ADK can’t constrain the output to a schema or stream JSON yet, so validate output in your app, for example with zod.

Options

temperature, maxOutputTokens, topP and topK work as usual:

const { text } = await generateText({
  model: adk(),
  prompt: 'Write a creative story',
  temperature: 0.8,
  maxOutputTokens: 500,
  topP: 0.9,
  topK: 40,
})