It works, I've shipped this as a "local inference"/poor person's ollama for low-end llm tasks like search. The main win is that it's free and privacy preserving, and (mostly) transparent to users in that they don't have to do anything, which is great for giving non-technical users local inference without making them do scary native things.
But keep in mind the actual experience for users is not great; the model download is orders of magnitude greater than downloading the browser itself, and something that needs to happen before you get your first token back. That's unfixable until operating systems start reliably shipping their own prebaked models that an API like this could plug into.
It's a one-time download shared by all websites that use the Prompt API.
What's a bigger issue is that the models on most standard PCs are both tiny and slow. I was going to try using the Prompt API to change the text of (infocom) text adventures on the fly. But for many PCs, this will currently be too slow to be feasible.
Seems like a good way for a rogue JS script to offload token generation to a bunch of unsuspecting visitors
It would actually be pretty interesting to see if its possible to decentralize the compute to generate something useful from a larger prompt broken down and sent to a bunch of browsers using a subagent pattern or something like RLM, each working on a smaller part of the prompt
Imagine a Vendor API that adds a way to link from the page straight into a device purchase workflow. As a trial of the API in Chrome you can order a new Google Pixel 9b directly from any page with the word Android in it!
Or a LocalNet API that integrates with trusted hardware devices on your local network. As a trial (Chrome beta programme — strictly limited but here’s 3x signup links to share with your friends) you can adjust your Google Next Mini underfloor heating directly from Chrome!
Or a DirectCast API that lets you stream <video> elements to a device of your choice even over a VPN. As a Chrome trial, you can use your Google Cloud account to stream directly from YouTube Premium to any linked Google Chromecast devices you own!
This API seems perfect for an idea I've had for a while: a de-snarkifier for social media.
Social media can be intellectually stimulating and educational, but it's also easy to get sucked into ideological sniping and flamewars, even if you didn't go looking for it. The emotional and intellectual energy spent flaming strangers on the Internet is a complete waste of human capital.
With an API like this, I assume you could have a browser extension that could de-snarkify content before showing it to you. You could ask the LLM to preserve all factual content from the post, but to de-claw any aggressive or snarky language. If you really wanted to have fun, you could ask it to turn anything written in an aggressive tone into something that sounds absurd or incompetent, so that the more aggressive the post, the more it would make the author look silly.
This could have a double benefit. For the reader, it insulates them from the personal attacks of random strangers on the Internet. Don't get me wrong, there is a time and a place for real, charged arguments about important issues that affect us all. But there is little to be gained from having those fights with strangers; on the contrary, I think it poisons the body politic when strangers are screaming at each other.
For the writer, it takes away any incentive to be snarky or rude. If other people filter their content this way, there's no point in trying to be mean to them, and no "race to the bottom" for who can be more nasty.
Or just ignore it. Or say you will not engage under [conditions]. Ultimately it will be you who looks foolish when the AI rewrote something incorrectly and you engaged with something that wasn't being said.
I would love an app like this. I am a frequent user of https://www.boringreport.org/ for news, which does something like what you’re describing but for news articles.
I was literally just thinking that I’d like something like this for HN, which has become an incredibly bitter, cynical, and depressing place in the last decade. On virtually any story, most of the top comments are negative. Every major company is a greedy monster trying to destroy your life, every CEO is a sociopath, everything is terrible, all the time. I wonder how most HN users even get out of bed every day.
I've thought about this for HN which, now that it's become so big, just has a lot of aggressive negativity and snark. You'd probably run into the same problem as Usenet Killfiles: the folks that use Killfiles would see random orphaned conversations or would just miss large parts of threads while the people that don't have Killfiles would see a mess of toxicity that would make them want to leave. Likewise if you prompt filter your experience, you'll be separating your experience from everyone else's.
I love this "de-snarkifier" idea and it seems to have broad interest. I couldn't resist hacking (well, vibe coding[1]) a "Snarknada" prototype to explore the viability, including patterns for low-latency and accuracy.
You’ve hit on exactly why we think on-device is the right move for this class of use cases. If you tried to "de-snark" an entire infinite-scrolling feed via a cloud API, the token costs would be astronomical for a developer. Plus, people (rightly) don't want to send their private social feeds or DMs to a third-party server just to clean up the tone.
Moving this to the device should make high-frequency "Semantic Mutation" financially and technically viable for the first time. If you (or anyone else) starts building this more seriously than my PM vibe coded toy, and hits specific friction points, I’d love to hear about them: it helps us prioritize the roadmap.
[1]: If you're using a coding agent (Cursor, Claude Code, etc.), I recommend pointing it to https://www.npmjs.com/package/built-in-ai-skills-md-agent-md. Most models were trained on the now-obsolete window.ai namespace, and this skill file helps them use the current APIs correctly.
been cranking on this too but not just for snark but for spam/scam heuristics too.
it's something I feel is finally viable to combat at zero cost to the user.
This plus webmcp would allow it to serve as a form of automod too on websites that you authenticate with (imagine a world where your social media profile has an automod of its own powered locally. can use this to steer your feed or to mute/block/moderate as needbe). Even without WebMCP I have been working on making it autodetect html elements and extract UGC (comments/threads..etc) automatically to moderate (since my initial tests with a small group found some websites with frequent UI changes would break if hardcoded or if they did a lot of AB testing)
Even better, the concept would allow you to also use it to hide certain spoilers (imagine sports or new movies that just came out and you want to not have to hide away from all socials).
didn't find any contacts on your new HN account, but in a few weeks will be able to reach out to you with it fleshed out. :)
We have a community of nearly 14k that we will distribute this to
I think it's a step into a future of proper Model API.
But it's just a small step.
It reminds me of Apple's Foundation Models [1]
While many AI integrations are focused on text communication / chat style.
A lot of software benefits from non-text interfaces.
I believe at some point OSes and browsers should provide an API to manage models so you'll have access to on-device/remote ones with a simplified interface for the app.
Making something standardized that is cross-platform would be fantastic. It also needs to be on mobile devices, so the players that can easily make it happen are mostly Apple and Google.
(Meta will follow or vice-versa I guess)
Key-point: it shouldn't be exclusive to promoted models.
The better part of this is having a local-first AI, particularly because it has tool-calling builtin & structured output.
I haven't pushed out a full version[1] which uses ducklake-wasm + this to make a completely local SQL answering machine, but for now all it does is retype prompts in the browser.
It's a tiny script that looks up the rss feed and uses the content to generate summaries; quite a nice fit with our static site. Sometime I'd like to extend it to ask different questions about the content.
This looks like it uses Gemini Nano under the hood. But the latest Gemma4 E2B and E4B models appear to be much better, so you'd probably be better off deploying quantized versions through an extension for now.
The idea of having local LLMs accessible in the browser for privacy concerning is nice i guess but when each browser has a different model attached to this API testing becomes even more a nightmare then now. I wonder if this will drive more users towards chrome because most of the usages of this API might be just tailored to fit the Gemini Nano model?
53 comments
[ 5.2 ms ] story [ 72.7 ms ] threadhttps://chromestatus.com/feature/6325545693478912
The parameters are not part of this initial release but can be added back with the origin trial you discovered.
But keep in mind the actual experience for users is not great; the model download is orders of magnitude greater than downloading the browser itself, and something that needs to happen before you get your first token back. That's unfixable until operating systems start reliably shipping their own prebaked models that an API like this could plug into.
What's a bigger issue is that the models on most standard PCs are both tiny and slow. I was going to try using the Prompt API to change the text of (infocom) text adventures on the fly. But for many PCs, this will currently be too slow to be feasible.
It would actually be pretty interesting to see if its possible to decentralize the compute to generate something useful from a larger prompt broken down and sent to a bunch of browsers using a subagent pattern or something like RLM, each working on a smaller part of the prompt
If you want to do anything interesting you need transformers.js and a decent mode. Qwen 0.9B is where things start working usefully
Or a LocalNet API that integrates with trusted hardware devices on your local network. As a trial (Chrome beta programme — strictly limited but here’s 3x signup links to share with your friends) you can adjust your Google Next Mini underfloor heating directly from Chrome!
Or a DirectCast API that lets you stream <video> elements to a device of your choice even over a VPN. As a Chrome trial, you can use your Google Cloud account to stream directly from YouTube Premium to any linked Google Chromecast devices you own!
Social media can be intellectually stimulating and educational, but it's also easy to get sucked into ideological sniping and flamewars, even if you didn't go looking for it. The emotional and intellectual energy spent flaming strangers on the Internet is a complete waste of human capital.
With an API like this, I assume you could have a browser extension that could de-snarkify content before showing it to you. You could ask the LLM to preserve all factual content from the post, but to de-claw any aggressive or snarky language. If you really wanted to have fun, you could ask it to turn anything written in an aggressive tone into something that sounds absurd or incompetent, so that the more aggressive the post, the more it would make the author look silly.
This could have a double benefit. For the reader, it insulates them from the personal attacks of random strangers on the Internet. Don't get me wrong, there is a time and a place for real, charged arguments about important issues that affect us all. But there is little to be gained from having those fights with strangers; on the contrary, I think it poisons the body politic when strangers are screaming at each other.
For the writer, it takes away any incentive to be snarky or rude. If other people filter their content this way, there's no point in trying to be mean to them, and no "race to the bottom" for who can be more nasty.
I love this "de-snarkifier" idea and it seems to have broad interest. I couldn't resist hacking (well, vibe coding[1]) a "Snarknada" prototype to explore the viability, including patterns for low-latency and accuracy.
You’ve hit on exactly why we think on-device is the right move for this class of use cases. If you tried to "de-snark" an entire infinite-scrolling feed via a cloud API, the token costs would be astronomical for a developer. Plus, people (rightly) don't want to send their private social feeds or DMs to a third-party server just to clean up the tone.
Moving this to the device should make high-frequency "Semantic Mutation" financially and technically viable for the first time. If you (or anyone else) starts building this more seriously than my PM vibe coded toy, and hits specific friction points, I’d love to hear about them: it helps us prioritize the roadmap.
[1]: If you're using a coding agent (Cursor, Claude Code, etc.), I recommend pointing it to https://www.npmjs.com/package/built-in-ai-skills-md-agent-md. Most models were trained on the now-obsolete window.ai namespace, and this skill file helps them use the current APIs correctly.
it's something I feel is finally viable to combat at zero cost to the user.
This plus webmcp would allow it to serve as a form of automod too on websites that you authenticate with (imagine a world where your social media profile has an automod of its own powered locally. can use this to steer your feed or to mute/block/moderate as needbe). Even without WebMCP I have been working on making it autodetect html elements and extract UGC (comments/threads..etc) automatically to moderate (since my initial tests with a small group found some websites with frequent UI changes would break if hardcoded or if they did a lot of AB testing)
Even better, the concept would allow you to also use it to hide certain spoilers (imagine sports or new movies that just came out and you want to not have to hide away from all socials).
didn't find any contacts on your new HN account, but in a few weeks will be able to reach out to you with it fleshed out. :)
We have a community of nearly 14k that we will distribute this to
While many AI integrations are focused on text communication / chat style. A lot of software benefits from non-text interfaces.
I believe at some point OSes and browsers should provide an API to manage models so you'll have access to on-device/remote ones with a simplified interface for the app. Making something standardized that is cross-platform would be fantastic. It also needs to be on mobile devices, so the players that can easily make it happen are mostly Apple and Google. (Meta will follow or vice-versa I guess)
Key-point: it shouldn't be exclusive to promoted models.
(1) https://developer.apple.com/documentation/foundationmodels So the app would be able to query and get the right model(s).
I haven't pushed out a full version[1] which uses ducklake-wasm + this to make a completely local SQL answering machine, but for now all it does is retype prompts in the browser.
[1] - https://notmysock.org/code/voice-gemini-prompt.html
It's a tiny script that looks up the rss feed and uses the content to generate summaries; quite a nice fit with our static site. Sometime I'd like to extend it to ask different questions about the content.
- Gemini Nano-1: 46% MMLU, 1.8B
- Gemini Nano-2: 56% MMLU, 3.25B
- Gemma4 E2B: 60.0% MMLU, 2.3B
- Gemma4 E4B: 69.4% MMLU, 4.5B
Sources:
- https://huggingface.co/google/gemma-4-E2B-it
- https://android-developers.googleblog.com/2024/10/gemini-nan...
Yes; "With the Prompt API, you can send natural language requests to Gemini Nano in the browser."