Models, models, models

Every month someone asks which model they should be using, and the honest answer is "it depends". So this time we're doing the whole spread in one night.

What to expect

• Shareouts — short, informal demos from whoever wants the floor. Local models, hosted APIs, the setup you're quietly embarrassed by. All welcome.
• Hands-on time — bring a laptop. Compare what actually runs well on your hardware against what you're paying for in the cloud.
• The usual — snacks, questions, and people happy to help you get unstuck.

Who it's for

Anyone curious about AI models, whether you've shipped something or you're still on your first prompt. Come excited and engaged — that's the only requirement.

Summary

Thanks to everyone who joined our 9th Vibe Coders Tokyo meetup! A huge thanks to Le Wagon Tokyo for hosting us and to all the speakers who shared their incredible experiences with models today. 🤩 ✨ Look out for our tenth event in September!

Demos

Your projects were absolutely wild!

• 📱 LLM on phone: Jon Gaul showed us how to scrape sites, run sub-agents, and manage files on a commute using Termux and Tailscale instead of standard AI apps.

• 🌐 Weblingo: Francois demonstrated a tool that dynamically translates complex SPAs into multiple languages by understanding the full page intent.

• 🦖 3D destruction: Ricardo Cabello blew our minds by using Cesium 3D tiles to build a Katamari Damacy-style game and a burning Shinjuku simulator running entirely in the browser!

• 🤖 Token logic: Shin broke down the real cost of tokens and when to use light, medium, or deep reasoning depending on the complexity of your coding tasks.

• 🔬 Local deep research: Stefania Druga shared an awesome resource for running deep research locally.

• 🌊 Nankai Trough: Marsel Bait shared an incredible mapping project visualizing the Nankai Trough.

Lastly, we scored 9/10 on the guess-the-designer.replit.app game! Well done team 👊