RSS feed
Subscribe to the technical notes feed.
This page is the readable version of the feed. The raw RSS XML is at https://blog.ryanspice.com/rss.xml.
Latest items
Recent feed entries
The feed includes the latest published public articles.
Why I Still Use Luna When Astra Is Available
A fast model is not a lesser model when the task is small, the feedback loop is tight, and token efficiency matters more than maximum reasoning depth. This is how I used GLM for the design direction and Luna for the last mile of a homepage redesign.
Building a Free Coding Fleet: MiMo Delegates, OpenRouter Free Lanes, and the Ox Alpha Discovery
How I built a zero-cost coding-agent fleet using MiMo Pro delegates, OpenRouter free-tier lanes, and the mystery Ox Alpha model that turned out to be GLM-5.3-Flash — now wired into Fugu and T3.
DeepSeek V4 Flash 0731: The Fast Lane Beside GLM-5.2 and Kimi K3
DeepSeek's latest public API release is V4 Flash 0731. Here is what changed, how its V4 Pro sibling compares with GLM-5.2 and Kimi K3, and why the benchmark chart needs a cost-and-harness warning label.
My Fugu/Fusion Harness Now Has One Conductor and Three Real Reviewers
I rebuilt my Fugu/Fusion coding harness around the models that are actually available: GPT-5.6 Sol conducts, Fugu executes through bounded specialist lanes, and a three-model remote Fusion panel reviews important work before the conductor decides.
PixelBoats Morning Watch: Lab-route proof without engine drift
Today's Pulse Pro focus is Lab routes: prove the composition-spike path, crosscheck gates, and keep parked engine-eval rows from becoming false P0 work.