Browser AI and Baseline 2026

Chrome is developing task APIs for built-in AI: Summarizer, Writer, Rewriter, and Translator, while the Prompt API offers a more general interface. The browser manages the model and may run it on the user’s device.

Useful frontend cases include article summaries, title generation, translation, alt text, and comment classification. AI becomes progressive enhancement rather than a required server dependency.

Support depends on the browser, device, and model availability. The interface must work without AI and improve when the API exists. Unsupported clients can use a manual flow or an explicit cloud fallback.

if ('Summarizer' in self) {
  const summarizer = await Summarizer.create();
  summary.textContent = await summarizer.summarize(article);
}

Expose model download and preparation states, with progress, cancellation, and a way to continue.

Baseline asks whether a feature is broadly available enough for ordinary use. The 2026 set includes container style queries, @scope, contrast-color(), active view transitions, the Navigation API, Trusted Types, and more.

Newly Available still does not mean every old device is supported; the product’s browser matrix remains the contract.

1. Define the product’s minimum browsers.
2. Check the feature in Baseline and compatibility tables.
3. Use feature detection, not user-agent sniffing.
4. Build a base flow without the new capability.
5. Measure polyfill cost against its value.

The best outcome is simpler code and a fast interface that degrades correctly.