46 entries · Checked 2026-10-10

YouTube channels

46 channels listed

gpu-hardware 16,900,000

Linus Tech Tips

@LinusTechTips

Consumer hardware at scale: GPU reviews and VRAM explainers, plus the server builds that follow them.

subscribers · read 2026-10-06
self-hosting 5,430,000

NetworkChuck

@NetworkChuck

Self-hosting, homelab and IT videos: servers and services you can run yourself.

subscribers · read 2026-10-06
gpu-hardware 2,720,000

Gamers Nexus

@GamersNexus

Independent GPU reviews and teardowns, including the power and thermal measurements that decide a card's place in a rig left on.

subscribers · read 2026-10-06
model-internals 1,720,000

Andrej Karpathy

@AndrejKarpathy

Builds and runs language models from scratch: the internals you need to reason about what can run locally.

subscribers · read 2026-10-06
local-models 1,440,000

sentdex

@sentdex

Long-running Python and machine-learning channel; builds and runs models on his own hardware and shows the benchmarks.

subscribers · read 2026-10-06
gpu-hardware 1,170,000

Hardware Unboxed

@HardwareUnboxed

GPU benchmarks and price-to-performance analysis: what a card does at a given power limit.

subscribers · read 2026-10-06
gpu-hardware 1,090,000

Jeff Geerling

@JeffGeerling

Tests GPUs, boards and clusters to see what runs on them, and publishes the numbers.

subscribers · read 2026-10-06
gpu-hardware 1,050,000

ServeTheHome

@ServeTheHomeVideo

Server and datacentre hardware reviews: motherboards, GPUs, power and storage.

subscribers · read 2026-10-06
local-models 755,000

Fahd Mirza

@fahdmirza

Daily hands-on videos on running and serving models yourself: Ollama, llama.cpp, vLLM, Open WebUI and RAG stacks.

subscribers · read 2026-10-06
local-models 638,000

Matthew Berman

@matthew_berman

Tests new models as they land, open-weight and local ones included, and says what they run on.

subscribers · read 2026-10-06
local-models 571,000

Alex Ziskind

@AZisk

Benchmarks LLMs and image models on consumer GPUs, Macs and mini-PCs, with the hardware price next to the result.

subscribers · read 2026-10-06
gpu-hardware 548,000

Level1Techs

@Level1Techs

Hardware and Linux in depth: multi-GPU workstations, server platforms, ROCm and NVIDIA drivers.

subscribers · read 2026-10-06
gpu-hardware 502,000

der8auer

@der8auer

GPU teardowns, power-limit and thermal testing: what a card does when you push it or cap it.

subscribers · read 2026-10-06
local-models 486,000

The AI Advantage

@aiadvantage

Practical AI tutorials and weekly updates, with regular local-model walkthroughs.

subscribers · read 2026-10-06
self-hosting 485,000

Hardware Haven

@HardwareHaven

Home servers, mini-PCs and self-hosted services; old and small hardware put back to work.

subscribers · read 2026-10-06
self-hosting 380,000

Craft Computing

@CraftComputing

Homelab and server builds and tutorials, following second-hand enterprise hardware.

subscribers · read 2026-10-06
self-hosting 338,000

Techno Tim

@TechnoTim

Proxmox, Kubernetes and self-hosted services at home: the infrastructure an inference box sits on.

subscribers · read 2026-10-06
model-internals 332,000

Yannic Kilcher

@YannicKilcher

Paper breakdowns and commentary on the open-model world, quantisation and local training runs.

subscribers · read 2026-10-06
gpu-hardware 323,000

Wolfgang's Channel

@WolfgangsChannel

Hardware and Linux reviews, including GPU and CPU buying advice.

subscribers · read 2026-10-06
local-image-video 304,000

MattVidPro

@mattvidpro

Tests local image and video models and the tools around them, and compares them side by side.

subscribers · read 2026-10-06
self-hosting 284,000

TechHut

@TechHut

Guides and tutorials on Linux, self-hosting and home servers.

subscribers · read 2026-10-06
self-hosting 277,000

Christian Lempa

@ChristianLempa

Homelab setup and infrastructure as code: running and exposing self-hosted services properly.

subscribers · read 2026-10-06
model-internals 253,000

Daniel Bourke

@mrdbourke

Teaches machine learning engineering, from the first model to running it on your own hardware.

subscribers · read 2026-10-06
local-image-video 250,000

Olivio Sarikas

@OlivioSarikas

ComfyUI, Stable Diffusion and Flux tutorials, with the workflows published.

subscribers · read 2026-10-06
local-models 235,000

bycloud

@bycloudAI

Frontier AI research explained, with regular tests of open models and the hardware that runs them.

subscribers · read 2026-10-06
local-models 226,000

All About AI

@AllAboutAI

Hands-on tests of new models and tools, including local models on his own GPU.

subscribers · read 2026-10-06
gpu-hardware 196,000

Actually Hardcore Overclocking

@ActuallyHardcoreOverclocking

Graphics-card PCB, VRM and power-connector analysis, down to the parts that decide a long run.

subscribers · read 2026-10-06
local-image-video 196,000

Theoretically Media

@TheoreticallyMedia

Puts the creative AI stack to a real test: local video, image and audio models, with the workflow and the hardware it needed.

subscribers · read 2026-10-06
local-image-video 182,000

Sebastian Kamph

@SebastianKamph

Guides on Stable Diffusion, Flux, ComfyUI and Forge from a working art director.

subscribers · read 2026-10-06
self-hosting 169,000

Awesome Open Source

@AwesomeOpenSource

Reviews of self-hosted open-source applications, from media servers to AI tools.

subscribers · read 2026-10-06
gpu-hardware 150,000

TechTechPotato

@TechTechPotato

Chip-industry analysis by Dr Ian Cutress: silicon, HBM and packaging, and the supply behind each GPU generation.

subscribers · read 2026-10-06
local-models 147,000

Hugging Face

@huggingface

The platform most open weights come from: model announcements, walkthroughs and tutorials.

subscribers · read 2026-10-06
local-models 135,000

Sam Witteveen

@samwitteveenai

Practical LLM development in code: agents, fine-tuning and running open models.

subscribers · read 2026-10-06
local-models 111,000

1littlecoder

@1littlecoder

Short, code-first videos on new models and frameworks, and how to run them yourself.

subscribers · read 2026-10-06
self-hosting 111,000

VirtualizationHowto

@VirtualizationHowto

Virtualisation, containers and home-lab software explained simply.

subscribers · read 2026-10-06
self-hosting 109,000

DB Tech

@DBTechYT

Self-hosting and home-server walkthroughs, including the AI services worth running at home.

subscribers · read 2026-10-06
gpu-hardware 103,000

Digital Spaceport

@DigitalSpaceport

Multi-GPU home servers in a garage datacentre: builds, parts lists, power draw and local model benchmarks.

subscribers · read 2026-10-06
local-models 96,200

Matt Williams (Technovangelist)

@Technovangelist

A founding maintainer of Ollama; walks through local model serving, evaluation and the tooling around it.

subscribers · read 2026-10-06
local-models 89,500

aiwithbrandon

@AIwithBrandon

Builds full-stack AI applications and shows the stack behind them.

subscribers · read 2026-10-06
self-hosting 83,900

Jim's Garage

@Jims-Garage

Homelab builds and home-server tutorials: Proxmox, storage and networking.

subscribers · read 2026-10-06
local-models 82,300

Mervin Praison

@MervinPraison

Tutorials on agents, retrieval and the models behind them, with the code.

subscribers · read 2026-10-06
local-models 77,600

Bijan Bowen

@BijanBowen

Tests new AI models and tools, local image and language models included, and shows the settings used.

subscribers · read 2026-10-06
local-image-video 66,300

Nerdy Rodent

@nerdyrodent

Open-source AI tutorials since 2020: local image models, and the code to run them.

subscribers · read 2026-10-06
local-image-video 38,600

Latent Vision

@LatentVision

Deep dives into diffusion samplers, schedulers and pipelines, mostly as Diffusers and ComfyUI experiments.

subscribers · read 2026-10-06
local-image-video 35,100

ComfyUI

@ComfyOrg

The project's own channel: the node editor most people use to run image, video and audio models locally.

subscribers · read 2026-10-06
local-models 26,300

Trelis Research

@TrelisResearch

Fine-tuning, inference and evaluation of open models, in long-form technical walkthroughs.

subscribers · read 2026-10-06

The number in each card's corner is the channel's subscriber count as YouTube reported it on the date printed underneath, read with yt-dlp from the channel's own page. No count is estimated, rounded down or carried over from an earlier reading, and a channel for which YouTube returned no count is not listed here at all. Every entry is a plain link: nothing on this page loads from YouTube, and nothing is embedded.

Why a channel list

Text is cheap and a demo is not. When a card, a quantisation format or a serving engine changes, watching someone else hit the wall first is the fastest way to know what you are buying. This page keeps the channels that do that work and nothing else. It is not a list of AI news accounts: a channel gets in only when running models on your own hardware is a substantial part of what it publishes, and the chip row says which of the five jobs it does.

How the list was built

  • The number in the corner of each card is the subscriber count YouTube reported on 2026-10-06, read from the channel’s own page with yt-dlp. It is the exact figure returned that day and the card prints the date underneath it. No count is estimated, rounded down, or carried over from an earlier reading.
  • No count, no entry. A channel whose page returned no subscriber count is left out rather than listed without one. A handle that resolved to a channel whose title did not match the one asked for — an idle or unrelated lookalike — was dropped as well, so every link here points at the channel the card names.
  • English only, judged from the text the tool returned: the channel title and description, plus the video titles for the few channels that publish no description. Text that was not English meant the channel was left out.
  • Subscriber count is the only comparable number YouTube publishes, not a quality score. The default order is by it because there is nothing better to sort by; a small channel that does the work is worth more here than a large one that talks around it.

What is deliberately not here

No embeds, no thumbnails, no player, no iframe: every entry is a plain link, so YouTube learns nothing about you until you click one. Nothing on this page loads from another site.