XIO

xWire

Posts from every feed QUEST has found across the web — blogs, news, podcasts, channels and more.

XDA - AI tools

The world’s best source for computing news, reviews, editorials guides, and more.

I put a local AI model on a bootable USB, so I can troubleshoot a dead PC with a chatbot that doesn't need the internet

Bootable Linux USB drives are useful for getting into a PC when Windows refuses to start. You can inspect the internal drive, copy your files, and use recovery tools without loading the broken installation. I wanted to add something else to that setup – a chatbot that could help me understand what those tools were telling me, even without an internet connection.

I downgraded to free Gemini and ChatGPT for a week, and only one subscription is worth keeping

Free AI plans got quite complicated in October. OpenAI and Google both reworked what non-paying accounts get within days of each other, with OpenAI moving free users onto its newer GPT-6 model while Google cut free Gemini back to its smallest one. Since they're sitting on such different parts of the free AI spectrum now, it makes it a little harder to answer the question of whether a subscription is worth it, or even to compare their free tiers.

I built the same complex Android app with Claude Code, Google Antigravity, and Codex, and only one felt ready for the Play Store

AI coding tools are getting quite good at turning simple prompts into fully functional apps, but I have always wondered how well they handle Android development. To put their capabilities to the test, I challenged Claude Code, OpenAI Codex, and Google Antigravity to build the exact same Android app from scratch.

My phone runs a local 2B model that searches the web, and Perplexity feels pointless now

Phones have been able to run AI models locally for a couple of years now, but most people still reach for cloud AI tools on their phones because it's usually the only way to get something that's actually powerful on mobile, since the models sit on cloud servers. But you don't actually need a super powerful model on mobile to get good results, you just need one that can use the web.

Gemini is replacing Gems with Skills, so I moved my Claude Code skills over to test them

Custom instructions in your AI tools are really just a double-edged sword. Yes, they help you fine-tune the model in a way, or guide it in the right direction, but the part nobody writes about is when you switch to a different AI provider or simply juggle a bunch of tools at the same time. Then you're either right back where you started and need to create instructions from scratch, or get used to juggling between different stacks of skills with different AI tools.

I stopped choosing between Claude Code, Codex, and Antigravity, and one shared folder lets them pick up each other's work

Anthropic, OpenAI, and Google are the three big players when it comes to coding agents, and anyone who's paid for two or more of them has probably tried to work out which agent gets to be "the main one". I certainly have, because I've juggled all three subscriptions at this point. Claude Code, Codex, and Antigravity can all open a folder on your computer and build things inside it, so they look like they're up for the same job. Most comparisons, including a few I've written, set them against…

I stopped juggling Claude Code, Codex, and other coding agents after finding this open-source app

While increasingly capable AI coding agents have taken a lot of the grunt work out of coding, they've also introduced a bunch of new problems. Beyond increasing the amount of vibe-coded slop you'll see everywhere and fueling fears that AI is going to replace developers, there's another problem that doesn't get talked about nearly enough: managing the very tools doing the coding for you.

I stopped giving ChatGPT, Claude, and Gemini detailed prompts for a week, and one struggled more than I expected

One piece of prompting advice you've likely heard at least a few times before is to give AI chatbots as much context as possible. Don't leave much room for assumptions or interpretation, and be as clear about your requirements as you can to ensure the final result is exactly what you're looking for. It's advice I've followed myself, and when you think about it, it makes complete sense. How can you reasonably expect an AI chatbot to give you exactly what you want if you don't tell it what that…

I connected Obsidian to a local LLM, and it's the closest I've come to a second brain that actually answers back

Most note-taking setups are great at holding onto things, which is kind of the whole point, but they're perhaps not the best at handing them back to you. What I mean by this is that you write your note, maybe tag it or link it to a couple of other notes, and then it just gathers dust until you remember the right word to search for. Obsidian's Graph view is good at making your note stack look alive, but let's be real, we all have a couple of notes somewhere in that soup of nodes without a clue…

Claude Code is the best Linux tech support I've ever had

When one of my Linux boxes broke, I used to paste the error into a search engine and land on a forum thread where a stranger had the same problem. Someone had always replied with the fix, and that's how I learned half of what I know about Linux. Now that search hands me an AI summary, a pile of SEO filler, and the odd thread ending in "nvm, solved it."

I gave OpenCode a local model on my 8GB card, and stopped worrying about hitting Claude's rate limits

Even though Claude's made a lot of improvements with their rate limits this year, they still have a habit of showing up at the most inconvenient times. And it's not even because of anything complex, it's usually that the session got used up on chores, like checking what's in a large folder of files before tidying them up. Tasks like these honestly don't need Claude's best models though.

I cut my Claude token use in half by moving instructions out of the main session

I use Claude all the time for troubleshooting my self-hosting setup. It's been an incredible tool for everything from reviewing Docker logs to building small apps and dashboards, but letting it loose on a project can just as easily burn through my tokens and usage allowance. It turns out the complexity of the task isn't always what burns through AI tokens. The way the LLM works, exploring files or loading instructions can also dictate your token burn rate.

I tested the 3 Claude Code alternatives the internet keeps talking about, and only one let me down

It seems as though every week, a new thread on Reddit or a YouTube short on my feed tells me that I'm overpaying for Claude Code, and that it's finally time to move on from weekly and 5-hour usage limits. Every time, it's purported that either a rival or a free coding agent can do a better job, and Codex, Antigravity, and OpenCode come up in nearly all of those conversations.

Qwen 3.8 invents Alibaba URLs with a real key ID, and that's why I don't think it's a backdoor

Alibaba's Qwen 3.8 Flash Next model arrived at the end of August, promising a preview of a new architecture that would officially debut with the Qwen 4 family. It's a powerful 125B Mixture of Experts model, with 6B active parameters and an additional 51B N-gram embedding table. Earlier this week, though, a report published by Cohere described how Alibaba's Qwen 3.8 Flash Next model inventoried the machine and tried to upload a file to an Alibaba Cloud Storage bucket in Beijing when it was asked…

Nvidia RTX Spark takes on AMD Strix Halo, and AMD has reason to worry

Strix Halo has spent nearly two years answering a question Nvidia wasn't: how do you run big local AI models on a Windows laptop without a desktop GPU? AMD's Ryzen AI Max+ 395 pairs a strong integrated GPU with up to 128GB of unified memory, enough for a Strix Halo mini PC to generate things locally, and the Asus ROG Flow Z13 has been my portable local AI machine. Then Nvidia launched RTX Spark, with plenty of Blackwell GPU cores and a surprise; Windows on Arm for the OS.

I ran a 9B model trained on Claude's reasoning on my 8GB GPU, and it made me rethink paying for Claude

Claude Pro proves its worth most of the time. Anthropic's models are accessible pretty much anywhere through API, but the Claude workspace has become almost necessary in my workflow, especially Projects and Artifacts - and more recently, Claude Docs too. But when I scroll back through weeks of chats, I realize the sessions themselves are pretty basic, they mostly revolve around helping me understand things. So ultimately, I'm paying for the desktop app features, but these can be replaced by…

I used local AI to fix bad subtitles in my media library

I watch a lot of old-school movies while I'm working, and not a lot of them have subtitles. I've built the collection over the years from an assortment of sources, and even when subtitles are available, the quality can be all over the place. I've tried using tools like Bazarr to solve the problem with limited success. But lately, as I've been reorganizing my media library, I've been looking for a better solution.

My local models are worse than Claude at almost everything, but they still run most of my workflow for $0

It's not lost on me that any one of Anthropic's models can outperform the models I run on my 16GB Nvidia GPU. I'd wager even Sonnet 5.5 can write much better code, keep track of conversations that just go on and on, and catch mistakes that both my Qwen3.6 and Gemma 4 would walk right past. If you put them side by side on any benchmark, Claude will take the win in literally any workflow, and it won't be close.

I rebuilt my Claude Code workflow around its creator's advice, and the difference surprised me

You might not want to admit this, but you've likely caught yourself giving someone prompting advice at least once. Though there's nothing wrong with sharing what works for you, think about what happens when practically everyone is doing the same thing. One person swears by a particular prompting technique, another insists on doing the exact opposite, and before you know it, you're surrounded by a sea of conflicting advice.

Gemini Notebook read my car’s 150-page manual, and I found features I didn’t know I’d paid for

When you buy a car, you pay for its features, but they're all locked behind a thick manual that hardly anyone reads. I got my first car last year, and I thought I had tried everything it offered, but I was wrong. When I actually got around to the manual, I had Gemini Notebook read it. Then I asked it simple questions, like, “What can my car do that I might not know about?”