Back to Xybrid

Live demo

The model runs here.

Your prompt. Your device. Try a small language model
on your GPU or CPU, right in this tab.

SmolLM2 135M
Runs on your device

Enter up to 280 characters. Use Command or Control and Enter to run.

Run with

Selects WebGPU when available, with a CPU fallback.

136 MB model download on first run.
Cached by your browser. Up to 256 output tokens.

02 Response

Ready when you are

A little intelligence.
Entirely on your side.

Choose a prompt, then run the model.

Your prompt stays in this tab.

No account. No API key. Inference happens here.

@xybrid/web
Build this into your appView the code
demo.ts
import { XybridLlm } from '@xybrid/web';

const llm = await XybridLlm.load(
  'https://huggingface.co/xybrid-ai/SmolLM2-135M-Instruct-LiteRTLM/resolve/main/model_metadata.json',
  {
    wasmPath: '/xybrid/runtime', // the runtime files, served from your origin
    accelerator: 'auto',
  },
);

const stream = llm.generateStream(
  "Write a haiku about AI that works offline.",
  { maxOutputTokens: 256 },
);
for await (const delta of stream) {
  // Append each delta to your response.
  console.log(delta);
}

Updates with your prompt and engine choice.

Explore the web SDK ↗

From this tab to your app.

Read the quickstart