VIBE CODERS COMMUNITY
← All events
Recap

Local Models with Gemma 4

WED, JUL 8 · GOOGLE SHIBUYA
Local Models with Gemma 4
Local modelsWorkshop

Gemma workshop, complete! 🎉 Thank you to everyone who braved the long lines to completely pack the room for our local models session! More info about the workshop.

Takeaways

  • 💡 You do not need a massive GPU to start vibe coding with Gemma! You can prototype rapidly right inside Google AI Studio for free.

  • ⚡ When running local models, your hardware dictates your limits. Tools like LM Studio can help you see if a model fits your GPU memory before loading it.

  • ➕ Apple silicon users have a superpower here. Using MLX optimized models on an M-series Mac with high RAM lets you run incredibly smart models locally.

  • 🧠 Use frontier cloud models for complex reasoning and local models for simple tasks to drastically cut your API costs.

A massive shoutout to Ju-yeong and @Alastair for dropping so much knowledge, and to our amazing volunteers for keeping the event running smoothly. The energy in the room while everyone was hacking was absolutely incredible! 🤩 ✨

Photos

Local Models with Gemma 4 — photo 1Local Models with Gemma 4 — photo 2Local Models with Gemma 4 — photo 3Local Models with Gemma 4 — photo 4Local Models with Gemma 4 — photo 5Local Models with Gemma 4 — photo 6Local Models with Gemma 4 — photo 7Local Models with Gemma 4 — photo 8Local Models with Gemma 4 — photo 9Local Models with Gemma 4 — photo 10
← Back to all events