Gemma 4 wows, Qwen still wins context

Gemma 4 looks like a real step up for open models, but this thread shows the local-LLM bar has moved from “can it run?” to “how much context can I actually afford?”

• –Google is pushing Gemma 4 as a serious open family, with sizes aimed at phones, laptops, and workstations rather than just datacenter use.
• –The praise here is real, but so is the tradeoff: Qwen still seems to offer better practical context windows and fewer hardware headaches for many local setups.
• –That makes context efficiency a first-class product metric, not a niche benchmarking footnote, especially for consumer GPUs and long chats.
• –Gemma 4 may still win for multilingual quality, writing style, and overall polish, which keeps it competitive even if it is less memory-friendly.
• –The launch reinforces a broader split in the open-model market: one track optimizes for capability, the other for usable scale on modest hardware.