Cutting · Engraving · Local AI

Laser and other tech projects,
always free for your use.

Every post has the details, process, and exact settings to duplicate it yourself. The site is live and maintained — and most of the work here, including this site itself, is done with LLMs.

A bear in a firefighter's helmet, laser-engraved into a curved acacia board

MakoBench

Language models scored on code and chat with one open test suite, run on our own hardware. Suite 4.1.

  1. GPT-6.2 Astra 25.4
  2. Claude Fable 5.2 25.2
  3. Kimi K3.1 Pro 24.6

See the full board →

Latest projects

Builds, guides, and experiments from the shop.

Tools & Downloads

WordPress to static site: the dead-simple kit, step by step

Move a WordPress site to a fast static site without being a developer and without typing commands: download our free kit, export your site, open the folder in the Claude desktop app, paste one prompt, check the preview, and put it on your web host. The kit was tested on a real WordPress site.

November 19, 2026
Tools & Downloads

Whole Site Export: your whole WordPress site in one zip

A free WordPress plugin that puts your whole site in one zip with one click: the full export file, every upload, every page as it really looks, and your settings. Built for slow shared hosting, tested in a real WordPress, with the limits stated.

November 19, 2026
Tools & Downloads

Tailscale Serve: a private web app on every device

Turn a small program on your own computer into a private HTTPS web app you can open, and install, on your phone and laptop with Tailscale Serve: no port forwarding, no public hosting, no monthly bill. Step by step, with a tested example app to download.

November 19, 2026
AI & Local LLM

GPT-6.2 benchmark: Astra and Sol scores and cost

GPT-6.2 Astra scores 25.4 on MakoBench 4.1, tied for first, with the best chat score on the board (13.7) and the fastest replies of the top group. Sol scores 23.9 for $1.04 a run, the cheapest of the leaders. Both failed one agent bug-fixing task that three rivals passed, and both do badly on explicit adult content.

November 18, 2026 · Updated November 18, 2026
AI & Local LLM

Model benchmarks, use cases and setup: November 2026

Model benchmarks, use cases and setup for November 2026: which of eleven models to pick for coding, chat or your own hardware, a five-step coding setup, and what a project costs. One standard run is $14.87 for Claude Fable 5.2 and $0.14 for DeepSeek V4.1-Flash-1017.

November 11, 2026 · Updated November 11, 2026
AI & Local LLM

Claude Fable 5.2 benchmark: where it wins, what it costs

Claude Fable 5.2 scores 25.2 on MakoBench 4.1, listed first in a statistical tie for first. It has the best chat score here (12.7), a perfect agentic pack, refused nothing, and costs $14.87 a run.

November 11, 2026 · Updated November 11, 2026
AI & Local LLM

MakoBench Suite History: Earlier Versions

The MakoBench tests change from time to time, and a score only means something inside the suite that produced it. Here is where the earlier boards live and why we kept them.

November 5, 2026 · Updated November 10, 2026
AI & Local LLM

MakoBench 4.1: What Changed

Suite 4.1 is a correction release with no new cases. Two independent model reviewers read every public answer key, about 190 problems were confirmed, and the board has been re-run: 31 of the 38 models scored on both boards rose, on both Chat and Code.

November 8, 2026 · Updated November 10, 2026
AI & Local LLM

MakoBench Suite 4.0: Nominal 20, Class and Lean

Suite 4.0 scores Code and Chat at 10 each instead of a percentage, drops the slowest and noisiest tests, and adds Class and Lean. Here is what the public release changed and what is in our own copy of the tool but not yet in a release.

November 5, 2026 · Updated November 10, 2026