Topic

Gemma

1 news article · 3 tutorials

News

Gemma 4 open model lineup, launches on AutoKaam

Gemma 4 Now Has Four Real Variants, E2B, E4B, 26B A4B MoE, and 31B Dense

Google's Gemma 4 lineup now lists four variants on official Google and Hugging Face pages: E2B, E4B, 26B A4B MoE, and 31B Dense. Audio is limited to the edge models, while the larger models are text and image models.

Aditya SharmaApr 10, 20268 MIN

Tutorials

Gemma 3 4B vision running locally on a GTX 1660 for screenshot OCR, automation tutorial on AutoKaam

AutomationLevel / intermediate

I Run Gemma 3 Vision On A 6GB GTX 1660 For Screenshot OCR: The Real VRAM And Latency Numbers

I host Gemma 3 4B vision on a single 6GB GTX 1660 for screenshot OCR and invoice extraction. Here are the install steps, the exact model tag, the VRAM it actually eats, and the cold versus warm latency I measured this week on my own desktop.

Published Jun 28, 20267 min read

Ollama running Gemma 4 9B in a terminal on Ubuntu

AutomationLevel / beginner

Ollama Gemma 4 On Linux, From Install To First Token

I ran Gemma 4 9B locally on a 32GB Linux box for a week of normal use. This is the Ollama install, the model variants worth pulling, and the one config tweak the docs do not flag.

Published May 6, 20266 min read

AutomationLevel / advanced

Running Gemma 4 Locally With Ollama, Setup Guide For Indian Devs

Local inference setup for Gemma 4 with Ollama on Mac, Windows, and Linux. Hardware requirements, performance benchmarks, and use cases where local beats cloud.

Published Apr 4, 20268 min read

Privacy choices