Is Ollama using your GPU?
Read ollama ps, confirm with nvidia-smi, fix a CPU fallback on Linux or force the CPU on purpose.
Arynwood tutorials
Learn a little.
Make it useful.
Start with Linux basics, get a local model running, or try a small browser project. These paths bring the existing Arynwood guides and tools into a practical order.
01 / Linux foundations
Get comfortable in the terminal
Look up a familiar task
Open the terminal study hub and search for files, disk usage or Git. Read the command’s explanation before trying it on your machine. The hub itself does not execute shell commands.
Compare related commands
Explore a category in the terminal study hub to compare commands grouped by purpose. Replace placeholders with your own values and pay attention to commands that change a system.
Plan a deployment
Use the deployment checklist as a reference for DNS, Nginx and publishing. It is a configurable checklist, not an automatic deployment service.
02 / Personal AI
Local AI questions, answered with measurements
How much VRAM does a local LLM need?
Measured memory for 3B to 33B models, how context length changes it, and what happens when a model doesn’t fit.
Open the VRAM answerWhat can an RTX 3060 12GB run?
Speeds for eight models on the card, from 3B to 33B, and which one to pick.
Open the RTX 3060 answerAre Mellum2.1 and Liquid’s d1 good for a local agent?
Seven models on an agent’s routing decisions, planted instructions and decision speed, measured on an RTX 3060.
Open the agent benchmarkDoes Ollama run faster on Linux?
The same used HP laptop on Windows 10 and Linux Mint 22.3: 9.4 against 12.9 tokens per second with the same model file, CPU only.
Open the laptop benchmarkCan you run a local LLM without a GPU?
CPU-only desktop and four-vCPU server speeds, an Ollama version comparison, and small-model tool tests.
Open the CPU benchmarkLocal AI on Linux
Arynwood MCP installation, hardware requirements and the optional services behind chat, knowledge search and creative tools.
Open Local AI on LinuxArynwood MCP demo
Explore the desktop interface in your browser using scripted conversations. No install required; this is not a live AI model.
Open Arynwood MCP demoStart with the hardware guide, install a model that fits your machine, and verify a simple chat before adding creative services. Need a hand? Explore paid setup help.
03 / Small creative projects
Sketch a workflow, then build a page
Map the idea
In the flowchart designer, add a Start/End node and a few Process nodes. Connect them in order, then export the Mermaid text to keep your diagram.
Preview a page
Open the HTML editor, change the starter heading and press Preview. Use Ctrl/Cmd + Enter to preview again after editing. Drafts stay in this browser’s local storage.
Try a design variation
The website prototype generator lets you explore a starter layout from design prompts. Treat its output as a starting point for your own edits.
04 / Take a break
Tune in and join the conversation
Listen to Jamendo radio
- Open the radio at the top of Arynwood Social.
- Choose a station preset, or use the genre menu and track search.
- Press play and adjust the volume. Playback starts when you choose it.
Visit the public chat rooms
Free AOL-style chat rooms in your browser: pick a nickname, press Connect and you’re in #public. No account needed.
Open Arynwood ChatWhat is IRC?
How Internet Relay Chat works, who still uses it, how it compares with Discord, and which app to use.
Open What is IRC?What happened to chat rooms?
Why AOL, Yahoo and MSN closed their rooms, what replaced them, and where chat rooms still exist.
Open the historyIRC beginner guide
Learn nicknames, channels, private messages and common IRC commands. Includes a separate operator reference.
Open IRC beginner guide