An independent research notebook on AI engineering, agents, models and the systems around them.https://julin.ai/atom/tags/performance/index.xmlJu Lin2026-10-31T00:00:00+13:00https://julin.ai/2026/10/31/small-python-v0-4-0/2026-10-31T00:00:00+13:002026-10-31T00:00:00+13:00<p>This is a follow-up to the <a href="https://www.julin.ai/2026/10/28/small-python-project/">Small Python</a> idea. I burned one more day of tokens to benchmark it and improve the performance. The setup is the same: one single Claude Code session, with <code>/remote-control</code> enabled so that I can prompt it while I&rsquo;m out. In the morning I attended a local farm event, and in the afternoon we went trick-or-treating for Halloween.</p> <p>The agent is capable of finding opportunities and fixing them correctly. Often it can run a few hours before it prompts me with more potential things to fix. I generally just ask it to continue.</p> <p>At one point it stopped and claimed it wouldn&rsquo;t gain much more, while it was still 1.3x–1.5x slower than CPython. I gave it a few ideas, and now it runs at 0.98x–1.1x of CPython.</p> <p>In the end, I&rsquo;m more than happy to ship a new version: <a href="https://github.com/soasme/small-python">Small Python v0.4.0</a>.</p>