posts
- Image input on a 27B coder costs 1.1 GiB, once
Image input for Qwen3.8-27B coding keys on an RTX 3090 cost 1.1 GiB at load on every variant. A full 121k-token context with a screenshot added 118 MiB.
- One open PR made MoE prompts 12-30% faster on my 3090
115 llama.cpp commits gave my RTX 3090 2.8% at most. One open PR from Unsloth's fork made MoE prompts 21-30% faster in llama-bench, 12-13% in llama-server.
- Three shutdowns, three causes
A storage box behind a DJI power station went down three times in two days with the battery near 95 %. The journals gave three different endings.
- The UPS said 2 minutes at 95 %
A UPS monitor took the minimum of three runtime estimates and raised low battery at 95 %. One of them was the station's time to full charge.
- The DJI went silent for four hours
A DJI Power 2000 stopped advertising over Bluetooth while the monitor served its last reading as live. Then the observer revived its own four-hour-old state.
- The screenshot opencode never sent
Local vision models on a Radeon 680M read test screenshots almost perfectly. Three opencode versions sent an error string instead until one config key was set.
- The leak scan found me in my own guard's tests
Publishing a private repo as one public commit: four gates, six catches, and my real user name inside my own guard's test cases.
- OpenCode didn't want an account. It wanted a model name.
A forwarded warning said OpenCode 2.x needs an opencode.ai account for self-hosted models. It doesn't. It needs you to name the model.
- My agent guardrails passed every test. Then I used them.
Weeks of tests against a scripted fake model, then one evening of real work with my coding-agent templates: three complaints, six findings.
- Eighteen days later, the same lie
A 122B model promised an atomic Java transfer it never wrote. Eighteen days later it did it again: all 4 new Java cells, 6 of 6 across both runs.