Search

Showing top 57 results for "Google Gemma XDA-Developers"

Google Gemma

Gemma is a family of open-weight language models released by Google for text generation and related NLP tasks.

146 articles indexed Last updated 17h ago See topic hub

Ollama is still the easiest way to start local LLMs, but it's the worst way to keep running them

…In my spare time, you’ll probably find me playing Counter-Strike or VALORANT, and you can reach out to me at adam@xda-developers.com , on Twitter as @AdamConwayIE , on Instagram…

Apr 8, 2026 · Adam Conway

I almost upgraded my GPU to run larger local LLMs, but this 8B model proved I didn't have to

…tiny Chromebook without dedicated GPU and barely any RAM). Google really ate with this one. I can even run the newer Gemma 4 12B on my PC when most guides will tell…

Jun 28, 2026 · Nolen Jonker

I ran Gemma 4 (26B) on a 10-year-old-GPU, and it's reliable enough to replace the cloud

xda-developers.com

May 10, 2026 · Ayush Pande

I added this open-source tool to my local AI stack, and my local LLM finally has persistent memory

xda-developers.com

Jun 9, 2026 · Nolen Jonker

I replaced the expensive Claude Pro subscription with these local models, and my productivity didn’t drop a bit

…Related I used Claude Code, Google Antigravity and OpenAI Codex to develop an app, and found only one worth using Vibe coding is here to stay, and it has only one champion…

Apr 21, 2026 · Parth Shah

I use Claude and local LLMs together now, and it costs half as much while being twice as fast

…For the local side of this pipeline, I went with Google's Gemma 4 26B model . It is highly capable, runs comfortably on my RTX 4070 Ti Super (notably, without the overhead…

May 23, 2026 · Abhinav Raj

I ran local AI models on a six-year-old laptop with no GPU, and they actually worked

…up Gemma 3 1B Easily the sweet spot for low-tier hardware machines Google's Gemma family tends to land in the sweet spot between capable and sluggish, and Gemma 3 1B…

Jun 5, 2026 · Samarveer Singh

The biggest local LLM on your machine is useless if it can't call a single tool, no matter how many parameters it has

…Google's Gemma 4 E2B, a 2.3 billion effective-parameter model with native function calling. It can even run on a phone . Google's own press release calls it "purpose-built…

Jun 10, 2026 · Adam Conway

My local LLM and Claude are helping me make my dream game, one day at a time

…a quantized version of Google's Gemma4 that I've managed to fit onto my GeForce RTX 4070 Ti. No, I didn't turn into a game developer through the use of…

Jun 14, 2026 · Samarveer Singh

I ran 14b and 20b local AI models on my laptop for the same tasks, and result was surprising

xda-developers.com

Jun 22, 2026 · Yash Patel

Followed topics

Search

Google Gemma

Ollama is still the easiest way to start local LLMs, but it's the worst way to keep running them

Top stories

Gemma 4's smallest model runs on 3GB of VRAM, and it's the one I actually reach for

I used Meta Llama 4, Qwen 3-Coder and Gemma 4 to develop a python app, and only one model is worth keeping for developers

I tested Google's new Gemma 4 12B on my 8GB GPU, and now I don't want to go back to smaller models

I tried Google's new DiffusionGemma, and watching it generate text like an image is unlike any local LLM

I almost upgraded my GPU to run larger local LLMs, but this 8B model proved I didn't have to

I ran Gemma 4 (26B) on a 10-year-old-GPU, and it's reliable enough to replace the cloud

I added this open-source tool to my local AI stack, and my local LLM finally has persistent memory

I replaced the expensive Claude Pro subscription with these local models, and my productivity didn’t drop a bit

I use Claude and local LLMs together now, and it costs half as much while being twice as fast

I ran local AI models on a six-year-old laptop with no GPU, and they actually worked

The biggest local LLM on your machine is useless if it can't call a single tool, no matter how many parameters it has

My local LLM and Claude are helping me make my dream game, one day at a time

I ran 14b and 20b local AI models on my laptop for the same tasks, and result was surprising