BLK · HOW-TO
How-to guides
0 task recipes. Each guide is designed around one job done end to end: steps, verification, and the failures that actually bite when you run open-weight AI on your own hardware.
Start here
Four recipes that turn reading into a working setup
Run one model
Pull with Ollama and verify a local response.
Open recipe →
Fix VRAM pressure
Reduce context instead of guessing at random settings.
Open recipe →
Measure tok/s
Turn your setup into comparable local evidence.
Open recipe →
Expose the runtime
Make Ollama reachable for apps and tools.
Open recipe →
Task finder
Find the recipe that matches the job
0 of 0 recipes
No matching recipe yet. Try another domain or open the benchmark request queue.
No how-to guides published yet.