Private AI
running inside your app.

# Stream a chat completion against the local server
curl http://127.0.0.1:11436/v1/chat/completions \
  -H "Authorization: Bearer $GEMMAKIT_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "gemma-4-e2b-it",
    "stream": true,
    "stream_options": { "include_usage": true },
    "messages": [
      { "role": "system", "content": "You are concise." },
      { "role": "user",   "content": "Summarise the changelog above." }
    ]
  }'

// Using the GemmaKit Swift client
import Foundation
import GemmaKitCore

let client = GemmaKitClient(
  baseURL: URL(string: "http://127.0.0.1:11436/v1")!,
  apiKey: ProcessInfo.processInfo.environment["GEMMAKIT_API_KEY"] ?? "local"
)

let stream = client.streamChunks(
  model: "gemma-4-e2b-it",
  messages: [
    .system("You are concise."),
    .user("Summarise the changelog above.")
  ],
  includeUsage: true
)

for try await chunk in stream {
  if let text = chunk.choices.first?.delta.content {
    print(text, terminator: "")
  }
}

// React Native, Electron, Ionic, Capacitor, Cordova, Tauri webviews, and Node
import { GemmaChatMessage, GemmaKitClient } from "@gemmakit/client";

const client = new GemmaKitClient({
  baseUrl: "http://127.0.0.1:11436/v1",
  apiKey: "local",
});

for await (const token of client.stream("gemma-4-e2b-it", [
  GemmaChatMessage.system("You are concise."),
  GemmaChatMessage.user("Summarise the changelog above."),
])) {
  console.log(token);
}

// Flutter and Dart applications
import 'package:gemmakit/gemmakit.dart';

final client = GemmaKitClient(
  baseUrl: Uri.parse('http://127.0.0.1:11436/v1'),
  apiKey: 'local',
);

await for (final token in client.stream(
  'gemma-4-e2b-it',
  [
    GemmaChatMessage.system('You are concise.'),
    GemmaChatMessage.user('Summarise the changelog above.'),
  ],
)) {
  print(token);
}

// Tauri backends and Rust applications
use gemmakit_client::{GemmaChatMessage, GemmaKitClient, GenerateOptions};

let client = GemmaKitClient::new(
    "http://127.0.0.1:11436/v1",
    "local",
)?;

let reply = client.generate(
    "gemma-4-e2b-it",
    vec![GemmaChatMessage::user("Summarise the changelog above.")],
    GenerateOptions::default(),
).await?;

# Reuse the official OpenAI Python client
import os
from openai import OpenAI

client = OpenAI(
    base_url="http://127.0.0.1:11436/v1",
    api_key=os.environ.get("GEMMAKIT_API_KEY", "local"),
)

stream = client.chat.completions.create(
    model="gemma-4-e2b-it",
    stream=True,
    stream_options={"include_usage": True},
    messages=[
        {"role": "system", "content": "You are concise."},
        {"role": "user",   "content": "Summarise the changelog above."},
    ],
)
for chunk in stream:
    print(chunk.choices[0].delta.content or "", end="")

// Reuse the official OpenAI TypeScript client
import OpenAI from "openai";

const client = new OpenAI({
  baseURL: "http://127.0.0.1:11436/v1",
  apiKey: process.env.GEMMAKIT_API_KEY ?? "local",
});

const stream = await client.chat.completions.create({
  model: "gemma-4-e2b-it",
  stream: true,
  stream_options: { include_usage: true },
  messages: [
    { role: "system", content: "You are concise." },
    { role: "user",   content: "Summarise the changelog above." },
  ],
});

for await (const chunk of stream) {
  process.stdout.write(chunk.choices[0].delta.content ?? "");
}