A hands-on local-model benchmark: porting an old screensaver to Python OpenGL with Qwen 3.8
prusswan · reddit · 2026-10-27
The author shares a benchmark task for gauging local model capability: use Claude Code as the harness driving a locally hosted Qwen 3.8 FN (Q4KXL via gufo on Strix Halo, Ubuntu 26.04) to port an old screensaver from the Wayback Machine to Python OpenGL with full visual fidelity — the comparison image shows matching startup frames.
Full run commands are provided, including an MTP speculative-decoding model, 133,760 context, and 32,768 max tokens. Some steps proved challenging within the 128K context window, but the verdict: the task is tractable on comparable hardware, and others are encouraged to reuse it.
More from coding & agent
- Open-source auto-research agent hits Top 2 on GPU kernel challenge in 7 hours — sguada · 2026-10-27
- 'Codex decided': devs admit making up reasons in code review for agent-written code — FlatStill2540 · 2026-10-27
- Running AI agents on a legacy IBM WMS: enforce tool order, script checks, verify evidence — ksrijith · 2026-10-27
- 365 public MCP servers tested: 51% lack tool input descriptions, the most common bug — abhiramreddykalluru · 2026-10-27
- 300 Agents, 60 People: Runlayer Found Engineer Details Real-World Agent Security Failures — AI Engineer · 2026-10-27
- CoreWeave ARIA agent ships four new live in-chat panels for ML run analysis — _ScottCondron · 2026-10-27