Qwen Ships Dense Model

HangzhouQwen’s new model handles 1 million tokens.

The open-weight release lets developers run vision-capable AI without per-token cloud fees.

The 27-billion-parameter model mixes cheaper linear attention with full attention, cutting the compute cost of million-token context.

A hosted Qwen Cloud version with built-in tools is coming soon, the company said.

Sources: Hacker News