Experience with Deepseek 4.1 Flash, current best free cloud chat model?

As the title says, I was having lots of issues by being a free user, google only provides the old reliable Gemini pro, but it has become so outdated and limited in use. It was nice for planning though, 7/10 work horse. GLM 5.3 seemed like a great promise but planning is very average, it loses track on longer projects and plans easily. It's good at coding but not unlimited use on it. GLM 5.2 has been my daily planning model, qwen 3.8 27B is the workhorse locally. I do consult Qwen max 3.8 but the results take so long in free chat and the responses are far too long, unreadable. So I went and tried DeepSeek 4.1 Flash, even after the Pro being worse debacle. I did like Flash locally it was very good but 4.1 is a huge step up. Very logical, identifies issues easily(multi and meta), very, very fast even for free use, explains clearly. It has been a joy to use, anyone else liked the output of the Flash 4.1 specifically for planning and reasoning tasks? It's not infallible, some small logical mistakes, but even Claude has done those so nothing new with LLMs: you always need to be in the drivers seat when making plans, so that it will follow a clear goal. I do really think we are starting to hit a wall with LLMs if a flash model is this good, yes it's huge, too big for my pc anymore, the old flash did fit.