HN Top 10 — July 19, 2026
Today’s Top 10 on Hacker News
1. Claude Code uses Bun written in Rust now
⭐ 52 💬 18 👤 tosh 🔗 Discuss on HN
This post likely discusses Anthropic’s Claude Code AI assistant adopting Bun as its JavaScript runtime, leveraging Bun’s Rust-based architecture to boost performance and execution speed. The discussion probably explores the technical benefits of this integration and how it streamlines the tool’s overall coding workflow.
2. Qwen 3.8 Max Preview
⭐ 26 💬 2 👤 lebovic 🔗 Discuss on HN
This post likely discusses an early preview of Alibaba’s Qwen 3.8 Max large language model, highlighting its updated architecture, performance benchmarks, and new capabilities for developers and researchers.
3. Qwen3.8 is launching and going open-weight soon
⭐ 205 💬 92 👤 nh43215rgb 🔗 Discuss on HN
This post likely discusses the upcoming release of Qwen3.8, a new large language model from Alibaba, which will soon be made available with open weights for public use and fine-tuning. The announcement is expected to spark discussion around its technical capabilities, licensing terms, and how it compares to other open-source AI models.
4. Transcribe.cpp
⭐ 550 💬 113 👤 sebjones 🔗 Discuss on HN
This post likely introduces an open-source C++ library for audio-to-text transcription, focusing on efficient native speech recognition. It probably covers the project’s architecture, performance benchmarks, and integration guidelines for developers building transcription features directly in C++.
5. Blender 5.2 LTS
⭐ 20 💬 7 👤 makizar 🔗 Discuss on HN
This post likely announces the release of Blender 5.2 LTS, a long-term support version of the open-source 3D creation suite designed for extended stability and professional workflows. The discussion will probably cover new features, performance optimizations, and the practical benefits of adopting a stable release for production environments.
6. Speech Recognition and TTS in less than 500kb
⭐ 467 💬 62 👤 petewarden 🔗 Discuss on HN
This post likely showcases a highly optimized software library or framework that delivers both speech recognition and text-to-speech capabilities in a remarkably compact package under 500 kilobytes. It probably explores advanced model compression, quantization, or efficient algorithm design to enable lightweight voice AI for bandwidth-constrained, mobile, or embedded environments.
7. OpenAI reduces Codex Model Context Size from 372k to 272k
⭐ 15 💬 1 👤 AmazingTurtle 🔗 Discuss on HN
This post covers OpenAI’s decision to shrink the maximum context window of its Codex AI model from 372,000 to 272,000 tokens. The reduction will likely impact developers who rely on the model to process large codebases or extensive documentation within a single prompt.
8. Half a Second – a book about the XZ backdoor
⭐ 12 💬 3 👤 zvr 🔗 Discuss on HN
This post highlights a new book titled Half a Second that chronicles the 2024 XZ Utils supply chain attack, detailing how a hidden backdoor was nearly embedded into a critical Linux compression library. The work likely examines the technical investigation, the rapid response by security researchers, and the broader implications for open source software security.
9. Better and Cheaper Than IPTV
⭐ 215 💬 53 👤 xonery 🔗 Discuss on HN
This post likely explores an alternative streaming solution or open-source project that delivers television content over the internet with superior reliability and lower costs than traditional IPTV services. It probably details technical setups, performance comparisons, and practical guidance for users seeking a more efficient way to stream live and on-demand video.
10. Codex Resets
⭐ 197 💬 139 👤 denysvitali 🔗 Discuss on HN
This post likely covers a discussion or release about resetting the context, state, or configuration of OpenAI’s Codex AI coding tool. It probably explores methods for clearing accumulated prompts, improving code generation reliability, or managing session limits in AI-assisted development workflows.