Pretty hype today: Gemma4 went open source under Apache 2.0, and Qwen 3.6-Plus dropped too.

Big day for AI news. Google open-sourced Gemma4 under Apache 2.0, which is super friendly for commercial use—basically you can just grab it and build products without worrying about hidden restrictions in the license, unlike some other open-source models. Over at Alibaba, Qwen 3.6-Plus also dropped, and damn, domestic large models are iterating fast. Also, OpenAI acquired the podcast platform TBPN, which looks like a move into content and distribution.

For those of us actually building stuff, Gemma4 going Apache 2.0 is the most interesting part—it means we’ve got another solid base model we can freely modify and use commercially. Qwen’s been updating regularly, and the Plus version usually brings better capabilities and longer context. Three big announcements in one day, that’s enough to keep us busy digesting for a while.

Yeah, Apache 2.0 is really nice for that, no guilt about using it commercially.

Gemma series has been pretty solid for local runs, waiting for the quantized version of 4.

Alibaba’s updates are seriously relentless, can’t keep up at all.