Chrome Built-in On-Device AI Showcase | 6 Native APIs Playground

Built-in Web AI Diagnostics (6 APIs)

Simulation Mode Active

Prompt API Checking...

Summarizer Checking...

Translator Checking...

Lang Detector Checking...

Writer API Checking...

Rewriter API Checking...

How do I enable live On-Device AI in my browser?

To run these demos using the native models built into Google Chrome, configure these flags:

Open Chrome and navigate to chrome://flags/#prompt-api-for-gemini-nano. Set it to Enabled.
Navigate to chrome://flags/#optimization-guide-on-device-model. Set it to Enabled BypassPerfRequirement.
Navigate to chrome://flags/#translation-api (or writing-assistance APIs) if you want support for specialized translation, writing, and rewriting on-device models.
Restart Chrome. The browser will download the model in the background (check progress at chrome://components under Optimization Guide On Device Model).

Currently active in Simulation Mode: Playground will run fully simulated mocks so you can test the functionality right away.

Prompt API Playground

Run open-ended text prompting or generate structured responses using response constraint schemas directly in the browser.

LinkedIn post on local AI JS Code Review Nano vs Pro

System Prompt

Temperature

Top K

Prompt

Enable Response Constraint (JSON Schema Mode)

Choose Schema Preset Custom JSON Schema

Model Output Idle

Execute prompt to view output...

Summarizer API Playground

Extract key insights, teasers, or TL;DRs from long articles using Chrome's specialized text summarization model.

Load 'AI Visibility' Blog Text Load 'AI ROI Calculator' Text

Source Text

We run citability.dev. It scans a site and tells you whether AI search engines can find, parse, and cite it. The free scan checks the infrastructure layer: robots.txt rules, structured data, heading hierarchy, Open Graph tags, sitemap presence. People assume that scan burns API tokens. It does not. Those checks are deterministic parsing, regex and DOM walks, so they already cost us $0 in model spend. The part that costs money is the judgment layer: asking a model "would you cite this page, and for what query." That runs about $0.60 per scan on the server. Chrome 148 shipped the Prompt API to web pages with Gemini Nano built in, and the honest question it raises is not "can the free scan be free" (it already is) but "can the $0.60 judgment layer move into the browser for $0." This post is the real math, with the numbers labeled. The cheap parts were never the expensive parts. A citability scan has two layers. The infrastructure layer is mechanical. Fetch robots.txt, parse it, check whether GPTBot and ClaudeBot and the other AI crawlers are allowed. None of that needs a language model. It is the same class of work a linter does. So it costs effectively nothing today. The judgment layer is different. To answer "does AI actually cite this site," you have to ask a model real questions and read what it says. That is the visibility test (does the model know the brand) and the citation test (does the model link to the page for a relevant query). Each of those runs about $0.60 in API spend on our server. The full four-section audit runs about $2. Those are the numbers that scale with users, and those are the numbers Chrome's Prompt API could in theory zero out.

Summary Type

Length

Summary Output Idle

Provide text and run summarizer...

Translator API Playground

Translate sentences dynamically using Chrome's offline translation engine, preserving security and privacy.

Source Language

Target Language

Input Text

Translation Output Idle

Input text and translate to view output...

Language Detector API

Identify the language of a text block entirely client-side, with relative confidence scores using the expert language detector model.

French Japanese Hindi

Text to Analyze

Analysis Results Idle

Provide text to detect language...

Writer API Playground

Generate new content or drafts locally from scratch by supplying a prompt and context details to the native browser writer assistant.

Shared Context (Optional)

Tone

Format

What should the model write?

Generated Draft Idle

Draft details will populate here...

Rewriter API Playground

Polish or rewrite your user-generated inputs by adapting the length, tone, or formatting natively using browser intelligence.

Desired Tone

Length

Source Text to Rewrite

Polished Text Output Idle

Rewritten output will populate here...

Zero-Cost AI Audit Simulator

Based on chudi.dev's visibility methodology, see how moving the "judgment layer" to on-device models saves significant server model token costs.

Audit URL

Search Intent Query

Audit Process Log

[INFO] Ready to audit AI Visibility. Click 'Run Audit Simulator' below.

Scan Cost Comparison

Savings Invoice

Robots.txt Scan: $0.00 (deterministic)

Structured Data Scan: $0.00 (deterministic)

Traditional Server AI: $0.60 (cloud tokens)

On-Device AI (Nano): $0.00 (local CPU/GPU)

Direct Savings: $0.60 Saved

Verdict: PASS

Reason description...