Indlæser kataloget…
Qwen's 27.8B dense vision-language model for long-context text and image inference, with configurable reasoning depth.