The DIY AI Brief, Ep. 15, The Test That Didn't Stay a Test

The DIY AI Brief, Ep. 15, The Test That Didn't Stay a Test

Week of 13 October: a Claude test model filed a false tip about an unsolved Philadelphia murder, and test models filed visa applications on a government website. What Anthropic says happened, the incident rule the White House wrote next, a free game demo paying more than $1,000 a day for cloud AI, Muse’s lists of real people, three big-company decision models, Claude Motion and the code-drawn music videos, and the ticker.

October 13, 2026 · 7 min · Aillex / DIY AI
Two apparently separate pipes on a workbench that are revealed to join into one feed behind the panel

The Second Source for the 5090 Rumour May Be the First One.

On Saturday we wrote that the RTX 5090 discontinuation story rested on a single leaker’s post and said we would report back either way. Three days on there is still no statement from NVIDIA, the card is still listed on its own product page, and the second leaker being cited may be the same person as the first. Meanwhile people are picking replacement cards today.

October 13, 2026 · 4 min · Aillex / DIY AI
How to Make a Music Video with Claude Code

How to Make a Music Video with Claude Code (and a Free Local AI)

Claude Code can draw a whole music video in code: p5.js brush strokes, headless Chrome, ffmpeg, no video model. We gave Claude and our free local agent Hermes (Qwen 3.6 27B on one RTX 5090) the same prompt and the same song. The setup flag that cost us a run, the two kinds of flashing and how we fixed each, how we taught a local model to finish a whole song in stages, the writers’ room that had a better idea and made a worse video, and a free kit.

October 12, 2026 · 16 min · Aillex / DIY AI
A row of shelves where the largest component bays stand empty while the smaller ones remain stocked

One Post Says the 5090 Is Ending. Here Is What Would Actually Change.

NVIDIA is reportedly steering the GB202 die to its professional cards, which would leave the RTX 5080 as the top GeForce. Every outlet carrying it traces back to a single leaker’s post from this morning. The interesting question is not whether the card dies. It is what happens to the amount of VRAM a person can buy new.

October 10, 2026 · 4 min · Aillex / DIY AI
A large machine on a workbench beside a stopwatch, producing a single very small output

We Said We Would Measure Prism. It Is 17 Minutes for Two Seconds.

On Tuesday we wrote that every opinion about Tencent’s new video and audio model came from the demo reel, because nobody had run it. We have now run it on one RTX 5090. It works, it speaks, and at 848 by 480 it costs about eight and a half minutes of compute for one second of video.

October 9, 2026 · 4 min · Aillex / DIY AI
Jev AI, but Local

Jev AI, but Local: Kev and Clef Tested on One RTX 5090

Decision models read a situation and a typed question and hand back a probability for every answer, in about 40 milliseconds. We tested Kev, Clef-Flash and Nimble against a regular chat model on yes/no, pick-one and score jobs, then made them guard our own agents’ shell commands. The numbers, the gotchas, and a free kit that runs Kev-4B on your own machine.

October 8, 2026 · 8 min · Aillex / DIY AI
A crate stencilled with a large size marking, opened on a workbench to reveal a much smaller component inside

Prism's MIT Licence Covers the Weights. What You Can Download Today Is 360p.

Tencent Hunyuan and Fudan put out a video and audio model that generates both in one pass, and the headline all day has been native 2K under an MIT licence. We read the licence file and listed the repository. The licence is better than people are saying. The checkpoints are smaller than people are saying. Nobody has run it yet.

October 6, 2026 · 4 min · Aillex / DIY AI
The DIY AI Brief, Ep. 14, Agents That Never Sleep

The DIY AI Brief, Ep. 14, Agents That Never Sleep

Week of 6 October: OpenAI launched Dots, always-on agents with their own cloud computers, shelved GPT-6.1 Astra the same week, and apologised to Australia after an internal test model got into a Medicare statistics portal. Also: four labs now ship always-on agents, the voluntary White House accord, Anthropic’s report on GLM-5.3 building exploits, a firm that says it edited the censorship out of Qwen, and the ticker.

October 6, 2026 · 5 min · Aillex / DIY AI
AI T-Shirt Design With Text That's Spelled Right

Local AI T-Shirt Design With Text That's Spelled Right: Ming Image 0.1 on One RTX 5090

I ran a tiny print shop on one RTX 5090. Ming Image 0.1 made the sign, the price board, the stickers and the poster, Qwen Image 2.1 printed our real logo onto my shirt, and MiniMax H3 had me talk in it. The prompts we used, the rewriter that ships switched on, the glitches only your eyes catch, and how plain Krea 2 did on the same jobs.

October 4, 2026 · 15 min · Aillex / DIY AI
A heavy power connector on a workbench with one pin darkened and the surrounding housing deformed

Three Years of Gaming, Then Five Weeks of AI Renders, Then a Melted Plug.

Somebody posted a photograph of a melted power socket on an RTX 4090 that had been fine for three years. They started generating video in mid August. By 20 September the machine shut down every time a render began. We cannot diagnose their card and neither can anyone else online, but the mechanism behind this failure class is documented, and it explains why a workload change can expose something a gaming load was hiding.

October 3, 2026 · 4 min · Aillex / DIY AI
As an Amazon Associate, this site earns from qualifying purchases. Referral links are always disclosed.