Jina AI has announced the release of two new Small Language Models (SLMs), Reader-LM-0.5B and Reader-LM-1.5B. These models, inspired by Jina Reader, are specifically designed to convert raw HTML into clean markdown efficiently. The Reader-LM models support multiple languages and can handle long contexts. They have outperformed other prominent models like GPT-4, Gemini-1.5, LLaMA-3.1-70B, and Qwen2-7B-Instruct in HTML-to-Markdown tasks. Reader-LM-0.5B has 494 million parameters, while Reader-LM-1.5B has 1.54 billion parameters. The models also excel in web data extraction and cleaning.