After a year of self-hosting LLMs, I realized the real bottleneck isn’t the GPU
…Sign in to your XDA account For the past year, I’ve been running my own local LLM setup, hoping it would make my work faster and more efficient. And in many…
Search
All sources
xda-developers.com
50
huggingface.co
40
developer.nvidia.com
12
intel.com
6
research.google
6
theregister.com
5
phoronix.com
3
techcrunch.com
3
spectrum.ieee.org
3
404media.co
2
amd.com
2
theverge.com
2
Tracked topic
LLMs
Large language models are machine learning models trained to predict and generate text and other language-based outputs.
Videos
More videos
Vector Search with LLMs - Computerphile
Nicholas Carlini - Black-hat LLMs | [un]prompted 2026
AgentMerge: Enhancing Battlefield Issue Management with LLMs | AI and Games Conference 2024
A Compilation of Robots Falling Down at the DARPA Robotics Challenge
DARPA Robotics Challenge 2013: A Woodstock for Robots | The New York Times
Top stories
I replaced cloud LLMs with local models running off a Proxmox LXC, and the performance trade-off was worth it
1d agoLLMs believe false statements even after explicit warnings that they're false
2d agoPaper page - Can LLMs Introspect? A Reality Check
3d agoMy self-hosted LLMs are a lot more than just a chat replacement – here's how they boost my productivity
5d agoI thought I needed a GPU for local LLMs until I tried this lean model
…Sign in to your XDA account When it comes to local LLMs, we have been told that if you aren’t packing a high-end GPU with a massive pool of VRAM…
Paper page - MinT: Managed Infrastructure for Training and Serving Millions of LLMs
…Managed Infrastructure for Training and Serving Millions of LLMs Published on May 13 Submitted by Andrew Chen on May 14 #1 Paper of the day Mind Lab Authors: , , , Andrew Chen , , , , , , , , , , , Nolan Ho…
I’d do these 5 things differently if I started self-hosting LLMs today
…Sign in to your XDA account I’ve been self-hosting LLMs for many months now, and the journey has been a wild mix of "aha!" moments and late-night troubleshooting sessions…
Discussions and forums
Measuring LLMs' ability to develop exploits
Measuring LLMs' ability to develop exploits
3
Measuring LLMs' ability to develop exploits
Measuring LLMs' ability to develop exploits
5
Ask HN: Are LLMs creating busy work?
When I look at what engineers and non-engineers are doing with LLMs (Claude Code and friends) at any company, I'm finding more and more instances of busywork.Burning tokens is equated with making progress. More conversat…
5
7
Is NVIDIA still the default best choice for local LLMs in 2026?
Is NVIDIA still the default best choice for local LLMs in 2026?
What rebuilding AlphaGo teaches us about self-play, RL, and future of LLMs [video]
What rebuilding AlphaGo teaches us about self-play, RL, and future of LLMs [video]
4
LLM Inference on Edge: A Fun and Easy Guide to run LLMs via React Native on your Phone!
LLM Inference on Edge: A Fun and Easy Guide to run LLMs via React Native on your Phone!
Optimize Fine-Tuning and Deployment of LLMs on an AI PC
…By Kelli Belcher AI software solutions engineer Fine-tuning and deploying large language models (LLMs) with billions of parameters requires significant memory and computational resources. To reduce these demands, we created a…
Post-Training Quantization of LLMs with NVIDIA NeMo and NVIDIA TensorRT Model Optimizer | NVIDIA Technical Blog
Agentic AI / Generative AI Post-Training Quantization of LLMs with NVIDIA NeMo and NVIDIA TensorRT Model Optimizer Sep 10, 2024 By Jan Lasek , Onur Yilmaz , Chenjie Luo and Chenhan Yu Discuss (0…
Google's Gemma 4 finally made me care about running local LLMs
…Our local LLM expert, Adam Conway, talked all about the nitty-gritty details in a separate article . But I do want to briefly explain how local LLMs actually work, because it makes…
Mastering Long Contexts in LLMs with KVPress
Mastering Long Contexts in LLMs with KVPress
Ollama is still the easiest way to start local LLMs, but it's the worst way to keep running them
…It's the Docker of local LLMs, and that comparison isn't a coincidence, given that some of the people behind Ollama came from the Docker world. Ollama's convenience comes at…