AI answers shouldn’t outgrow their interface
gpt_markdown 1.3.0 adds a faster single-pass parser, lazy sliver rendering, flat-cost streaming updates, animation, and a complete customization system.
A renderer designed for model output
gpt_markdown 1.3.0 is built for Markdown that arrives a few characters at a time, with incremental updates, streaming animation, and graceful handling of unfinished syntax.
Performance you can reproduce
The release includes a single-pass parser measured at 3 to 9 times faster, a lazy SliverGptMarkdown renderer, and published benchmark methodology with raw measurements and caveats.
Learn how Val approaches AI output rendering
