Honestly, before picking a specific model, I think its worth looking at what your actual performance requirements are. Extensions can get pretty laggy if the background scripts arent optimized, and some AIs tend to output really bloated code that might hit rate limits or slow down the browser. What kind of scale are you looking at for those external API calls? Also, is your priority more on getting the extension up and running as fast as possible, or are you looking for code that is strictly optimized for high-performance service worker execution?

> Also, how well do these models handle updates to Chrome's extension API? Ok so I actually spent some time running side-by-side tests because I was getting super frustrated with outdated code snippets and honestly it comes down to how they handle the context of Manifest V3 vs V2. I tested GPT-4o and Claude 3.5 Sonnet specifically on their ability to write a background service worker that doesn't crash on the first API call and tbh the performance gap was wild. Basically Claude 3.5 Sonnet was hitting a much higher success rate on the first try for complex async messaging while some other models were still trying to use 'background.scripts' which is totally deprecated now ur gonna find that some models are way better at reasoning through the newer V3 restrictions than others so it's not just about writing the code but about how many times you have to prompt it to fix its own mistakes because that's where the real time sink is and its honestly been a huge factor in my dev speed recently although I'm still not 100% sure why one is so much more consistent than the other lol.

This is exactly what I needed to hear. Youre a lifesaver honestly.

Hey there!

So, I'm kinda new to this too, but I've been playing around with Gemini for my Chrome extension project. Honestly, it's been pretty decent for generating basic manifest files and simple JavaScript snippets. I used it to create the event listener for my popup's button. It worked!

I haven't tried GPT-4 yet, but Gemini is a decent option to start with, especially if you're on a budget. Hope this helps! Feel free to ask questions; I'm learning too!

Hey there!

So, I see you're looking into AI for Chrome extension dev. Gemini's a good starting point, but honestly, don't sleep on GitHub Copilot. I've been using it for a few months, and it's been a game-changer, especially for JavaScript stuff! It's really good at suggesting code snippets, especially when dealing with event listeners and API calls. It's also pretty decent at understanding the Chrome extension API, although you still gotta double-check things, ya know?

One thing I really like is that it learns from your code style. So, the more you use it, the better it gets at predicting what you're trying to do. I think it's a paid service, but it's worth checking out the trial to see if it fits your workflow. It's saved me a ton of time, so I'm happy with it.

Hope this helps! Let me know if you need more info.

Hey! Something to consider, besides cost, is how accurate the AI is *specifically* with security stuff in extensions. Like, make sure it's not suggesting code that opens you up to XSS or something. Double-check everything, yeah?

Hey! I see you're trying to pick the best AI for Chrome extension dev. Gemini and Copilot are cool, but have you thought about the *cost* over time? I mean, those subscriptions can add up FAST!

Honestly, for generating snippets and manifest files, I've actually had decent luck with some of the *free* options. I'm talking about the free tiers of models like Bard or even just using simpler, open-source models locally. They might not be as fancy as GPT-4, but for basic stuff, they're surprisingly good!

Here's the thing: Chrome extension development often involves a lot of tweaking and debugging *after* the AI generates the code. So, spending a ton on a super-powerful model might not be worth it if you're gonna be editing things anyway. Start cheap, see what you can do, then upgrade only if you really need it! Good luck!

Hey! Jumping in here... While everyone's talking models, it's worth considering *where* those models are integrated. Gemini's great, but its code generation in the browser IDEs isn't always as smooth as dedicated tools. I've honestly had issues with some of the 'free' tiers of these AI coding assistants; they sometimes suggest really outdated practices, especially when it comes to Chrome's manifest version. Not ideal!

GitHub Copilot's integration within VS Code is, IMO, a more polished experience. It 'learns' your coding style better. But, and this is a big BUT, be super careful to validate *everything* it suggests against the official Chrome extension documentation. Security vulnerabilities are no joke. Also, remember that updates to the Chrome API can break things, and AI tools are only as good as their last training data. So, don't blindly trust anything! Consider a professional extension developer to double-check things, especially if you're handling user data. Hope this helps!

Hey! Everyone's talking models, which is cool, but honestly, have you considered *not* using an AI at all for the *entire* thing? I've built a few extensions, and sometimes the best approach is a hybrid. For manifest files especially, just use a template! There are tons online. UI? Same deal. AI is great for snippets, sure, but don't let it do *everything*. You'll learn more, and it might actually be faster in the long run! I totally get wanting to speed things up, but DIY-ing the structure can be super rewarding. Good luck!