#performance
The latest news with this tag.
News · #performance
All news with this tag →A new article compares JPEG XL and AVIF for web use-cases, discussing compression efficiency, decoding speed, and browser support. Developers can use the analysis to choose the most suitable format for reducing payload size while maintaining visual quality on their sites.
JPEG XL vs. AVIF: comparing web image formats ↗Firefox now supports resumable large file downloads, allowing interrupted transfers to continue without restarting. This change improves user experience for heavy content and can reduce bandwidth waste, which developers should consider when designing download flows or providing large assets.
Firefox adds resumable support for large downloads ↗A GitHub project recompiles PlayStation 2 games to WebAssembly, allowing them to run in browsers without native plugins. This showcases the growing capability of WebAssembly for complex, performance-critical workloads and may interest developers exploring high-performance web apps or game porting.
PS2 games recompiled to WebAssembly ↗A classic Quake engine has been rewritten in safe Rust and compiled to WebAssembly, allowing the game to run directly in a browser. This demonstrates Rust's safety guarantees and WebAssembly's capability for high-performance web apps, which may interest developers exploring low-level performance or alternative languages for the web platform.
Quake ported to safe Rust runs in the browser ↗Anthropic released Claude Haiku 5.5, a faster, lower-cost LLM that matches OpenAI's GPT-6 Luna pricing at $0.10 per 1 k input tokens and $0.50 per 1 k output tokens for up to 100 k tokens. It uses a less efficient tokenizer, increasing token count by about 25 % versus Haiku 4.5, and defaults to medium reasoning effort. Developers can test it via the llm-anthropic plugin and compare costs and performance for code-related generation tasks.
Claude Haiku 5.5 offers GPT-6 Luna-level pricing with higher benchmark scores ↗