Expect more benchmaxxed models from Meta. I wonder when will the AI community stop optimizing for lmsys arena.

The secret to scaling is always caching. h/t ⁦@hyperknot⁩

Join millions who have switched to Grok.

https://t.co/QluNK4JL4e

OpenFreeMap survived 100,000 requests per second I was about to post about how nice the last 10 months of OpenFreeMap have been. The architecture has really proven itself to be great, Cloudflare has agreed to sponsor the bandwidth, Hetzner servers are super stable as always,

Am I the only one who finds Sonnet 4.0 better at coding compared to Opus 4.1? Opus 4.1 is weird.

Kinda amazing: the mystery model "summit" with the prompt "create something I can paste into p5js that will startle me with its cleverness in creating something that invokes the control panel of a starship in the distant future" & "make it better" 2,351 lines of code. First time

Is CSS scrollbar-gutter support like totally broken in almost all browsers? This is Chrome. Safari is broken in a different way. I don't want to add JS based scrollbars, but it's really hard to fix this in a cross-platform way.

You'd never guess the new provider "market share" metric by @OpenRouterAI. Google is at 43%! The only problem is that it's not market share, but "token share". The Gemini Flash models totally skew this, as they are the cheapest large context models. @OpenRouterAI, it'd be

Ultimate marketing flex by @midjourney: 1. Make every user connect with the "Midjourney Bot." 2. Notify 21 million users of a new launch, for free! (Even if they left ages ago.)

unique product ideas in 2025 1. SEO articles generator 2. Reddit leads finder 3. Social media scheduling tool 4. Product Hunt killer directory 5. AI avatar reels generator why is nobody building them?

2 days ago I posted this screenshot about extremely fast responses on o3 models. I looked more into it, and it turns out @OpenRouterAI has buggy calculation for tps on o3 and o3 Pro models. If you divide total output tokens / total request time, you actually get reasonable

I bet OpenAI switched to a quantized model with the 80% price reduction.

I again had one of those moments, when I'm well into a 1000 word conversation with Sonnet (about deleting a file on macOS), and I'm getting stuck, it's starts repeating the same ideas. I ask o3, it spends 3 minutes thinking and gives me a single answer which immediately solves

One US client sent us a USD paper check in the mail, instead of the wire transfer / ACH we'd asked them to use. Now I have this piece of paper for $1100. What can I do with it? I'm in the EU and I have no idea where I can possibly cash or deposit this check. What else can I do?

Perplexity – which I thought would give the best answer as it can research online – gave a totally useless but super-long answer.

o3 also gave a nice answer, explaining that it's a change introduced in iOS 16:

Claude was absolutely wrong, recommending I use an Apple Watch and even saying that "Your 'apologize and call back' approach is probably the most reliable solution" :-)

OK, I asked Gemini 2.5 Pro, o3, Claude Sonnet 4, and Perplexity. Gemini gave a perfect answer! It explained why it happens: an accessibility setting was inverted on iOS 16, and many users got caught by this. I personally never touched this, so I'm surprised by its default value.

https://t.co/gmvhzQIS7X

Right now, you might have read the news that a Chinese paraglider pilot, Peng Yujiang, got sucked into a cloud and ended up at 8,598 meters / 28,000 feet. It's literally on every news website: CNN, BBC, The Guardian. Now, what makes this story really interesting is the