Divine Brings Back Vine, OpenAI Cuts the Emojis, and the Week the Internet Got Weird Again

Divine is Vine again, and the nostalgia wave just got official

Jack Dorsey's Divine app launched on the App Store this week, bringing back hundreds of thousands of videos from the original Vine platform alongside new uploads. If you were online between 2013 and 2016, you probably remember Vine as the birthplace of six-second comedy that launched whole careers. Twitter killed it in 2017, and people have been mad about it ever since.

Divine isn't just an archive. It's allowing new videos too, which means we're about to find out if the format still works or if the magic died with the original platform. Early reports say the app is functional and the old videos are there, but the real test is whether anyone under 25 cares about six-second loops when TikTok already owns short-form video.

Dorsey backing this makes sense—he co-founded Twitter, which owned Vine—but it's also a little strange to resurrect a format that lost to Instagram Stories and Snapchat before TikTok even existed. Maybe the constraint is the point. Maybe people are tired of 90-second skits. We'll know in a few months when the download numbers either spike or flatline.

OpenAI ships GPT-5.5 Instant and promises fewer emojis

OpenAI replaced the default ChatGPT model with GPT-5.5 Instant this week, and the main selling point is that it's "smarter and more accurate" while also dialing back the emoji spam. If you've used ChatGPT in the past year, you know exactly what they're talking about—responses peppered with 🎉 and 🚀 for no reason, like the model was trained on LinkedIn posts.

The update is live now, which means everyone using the free or Plus tier gets it by default. OpenAI didn't publish benchmarks or a detailed changelog, so we're left guessing what "smarter" means in practice. Faster reasoning? Better context handling? Fewer hallucinations? Probably all three, but the emoji fix is what people are celebrating.

There's also a separate ChatGPT app for iOS that's Codex-adjacent but not directly tied to the code-focused model. OpenAI is clearly segmenting the mobile experience—one app for general chat, one for code, maybe more to come. It's smart product strategy, but it also means you're juggling multiple apps for what used to be one service.

TikTok's "pinky time" trend is weird and probably pointless

TikTok creator Daniela Paez-Pumar started a trend where she and her friends do a finger sequence every night at 7:45 p.m.—middle and pointer fingers wrapped, ring fingers touching thumbs, pinkies wiggling up and down. The claim is that doing this daily can "slow down time" or help you be more present or something equally vague.

Does it work? Almost certainly not in any measurable way. It's a ritual, which means the benefit is whatever placebo effect you get from doing the same thing at the same time every day with people you care about. That's not nothing—rituals help with mindfulness and routine—but the specific finger choreography is doing zero biological work.

The trend is taking off anyway because it's low-effort and social. You don't need equipment, you don't need to believe in chakras, and you can film it for content. It's the wellness-industrial complex stripped down to pure vibes. If it gets people to check in with friends every night at 7:45, that's probably healthier than doomscrolling, but let's not pretend your pinky is accessing some hidden frequency.

iOS 27 will let you swap out Siri's AI backend

Buried in this week's Apple news is a report that iOS 27 will let users choose between Gemini, Claude, and other models for the system-level AI features currently powered by Apple Intelligence. That's a big shift—Apple has historically locked down anything related to the OS—but it makes sense if they want to avoid being stuck with one model vendor long-term.

The obvious win here is flexibility. If Google's Gemini gets better at summarization but Anthropic's Claude is better at reasoning, you can swap based on the task. The downside is fragmentation: now your iPhone's AI behavior depends on which model you picked, which means troubleshooting gets harder and the user experience becomes less predictable.

This also signals that Apple knows it can't win the foundational model race on its own. Letting users bring their own model is a smart hedge, but it's also an admission that the "Apple builds everything" strategy doesn't work for AI. We'll see if Google and Anthropic are willing to play nice or if Apple ends up with a worse version of each model because the best stuff stays exclusive to the native apps.