Flash 3.6 It seems to me that it is not an evolution
Ranking models
Ranking models in ai arena in text in creative writing is equal the power of the model in RP or its different and what's your opinions about top 10 of the list
Notebooklm for RP
​
Is anyone try notebook lm for role play, i think it's have a good potential, long memory 1 m tokens and good limits for gemini 3.1 pro (i think it's 500 message per day for pro user) , so is anyone try it can say his experience or Opinions ، When I first tried it, most of my problems with it were that it was too limited to the sources and couldn't create events or characters not found in the sources, plus it had to mention the rules written in the prompt every time it was applied.
Single character and scenario card
Throughout my life, I have never understood the concept of a single-character card. It's a short story where you know the ending, the beginning, and all the events that will occur in it. It's very trite: my aunt, my cousin, my friend, the traitor, the bully. Enough of this nonsense! Are there really people who enjoy this? Why is it that whenever I search for a card, all I find is this nonsense, and it's always on the trending page? Why is there so little content for scenario cards that tell a story or a world with an open map, which is the real enjoyment for people with brains?
مش طبيعي كمية التكتم المبالغ فيها الي عايزين يعملوها علي الحوار دا و مش مقتنع ان دا خوفا علي صحة الشعب المصري من نظام الطيبات رغم اني طبعا مش متفق معاه بس الفكرة في انه فيه خمسين الف واحد بيهري ب كلام زي ضياء و ممكن كمان اشهر منه و نظريات مؤامرة و كلام يودي في داهيا اكتر اشمعني يعني دا الوحيد الي يتخذو معاه الاجرائات دي كانهم بيدو ماتريل للناس الي متبعاه تحسسهم انه الصح و انهم مميزين انهم عارفين حاجة بيحاولو يخبوها عن باقي الناس او تخلي الموضوع غامض اكتر و الناس تحس ان فيه أن
Does anyone else feel like Gemini 3.1 Pro lacks deep understanding and is way more superficial and overly literal compared to Gemini 3.0 Pro or 2.5 Pro?
I always felt that when I talked to the older models—whether I was sharing personal problems or just giving them a complex prompt—they actually understood the intent behind my words and executed it perfectly. Now, it just follows instructions literally to a fault. Worse, it interprets human intent in a really weird, mechanical way.
I've given it so many chances, but the feeling remains. Even when I try to tweak or fix a prompt, it literally just copy-pastes and frankensteins my exact words together instead of grasping the underlying goal and rephrasing it properly like the older models naturally did.
Honestly, I'm getting really frustrated using it. It feels like Google completely prioritized coding capabilities and literal execution over deep comprehension and conversational flow for the general user—who, by the way, makes up a much bigger demographic than just programmers.
Note: excuse my language i make it with ai cause English not my first language