v2026.9.25.4
QwenLM/Qwen3-VLv2026.9.25.4Sep 26, 2026by synoet
Upgrading?
AI Summary
A minor patch release addressing stability issues with agents, specifically preventing the composer from growing incorrectly and keeping the last agent visible during loading.
Key Highlights
- Fixed the agent composer from growing from zero after an offscreen mount.
- Fixed agents disappearing while their availability is still loading.
- Fixed command routing to prevent commands from being sent to draining replicas.
Full Release Notes
## What's Changed * fix(agents): don't grow the composer from zero after an offscreen mount by @synoet in https://github.com/macro-inc/macro/pull/7030 * fix(agents): keep the last agent while its availability loads by @synoet in https://github.com/macro-inc/macro/pull/7029 * fix(agent_harness): route commands away from a draining replica by @404Wolf in https://github.com/macro-inc/macro/pull/7024 * fix(web): start feature flags from PostHog's cache by @synoet in https://github.com/macro-inc/macro/pull/7027 **Full Changelog**: https://github.com/macro-inc/macro/compare/v2026.9.25.3...v2026.9.25.4
Related Repositories
rasbt/LLMs-from-scratch★ 107,072
Implement a ChatGPT-like LLM in PyTorch from scratch, step by step
lyogavin/airllm★ 40,206
AirLLM 2.8T LLM inference with single 4GB GPU
NirDiamant/RAG_Techniques★ 29,960
This repository showcases various advanced techniques for Retrieval-Augmented Generation (RAG) systems. Each technique has a detailed notebook tutorial.
microsoft/OmniParser★ 25,539
A simple screen parsing tool towards pure vision based GUI agent
karpathy/nn-zero-to-hero★ 24,907
Neural Networks: Zero to Hero