This week in Python, the community discussed the use of Python's `__init_subclass__` over metaclasses and measures to prevent ZIP parser confusion attacks on Python package installers. A new large language model series, Qwen3, was developed by the Qwen team at Alibaba Cloud. Other highlights include the creation of an offline, open-source desktop Pixel Art Editor in Python and the representation of Python notebooks as dataflow graphs. In addition, there were discussions on Python performance myths and the development of a Python package, Llmswap, to reduce LLM API costs significantly. Wishing you a great week and happy coding!
Forget metaclasses; Python’s `__init_subclass__` is all you really need
PyPI: Preventing ZIP parser confusion attacks on Python package installers
QwenLM / Qwen3
Qwen3 is the large language model series developed by Qwen team, Alibaba Cloud.
Representing Python notebooks as dataflow graphs
I built an offline, open‑source desktop Pixel Art Editor in Python
Create space-saving clones on macOS with Python
Synchrotron, a real-time DSP engine in pure Python
Python performance myths and fairy tales
Llmswap – Python package to reduce LLM API costs by 50-90% with caching
joinly
Build personalized, conversational meeting agents in minutes
HRT's Python Fork: Leveraging PEP 690 for Faster Imports
Python Hub Weekly Digest for 2025-08-10
OpenPipe / ART
Agent Reinforcement Trainer: train multi-step agents for real-world tasks using GRPO. Give your agents on-the-job training. Reinforcement learning for Qwen2.5, Qwen3, Llama, Kimi, and more!
PYX: The next step in Python packaging
Why Python's deepcopy() is surprisingly slow (and better alternatives)
Project by Ruslan Keba. Since 2012. Powered by Python. Made in 🇺🇦Ukraine.