a 7M parameter model just beat DeepSeek R1’s 671 billion parameters on hard reasoning tasks. 45% vs 15.8%. trained in hours. fits in 28MB. runs on a single GPU. the trick is that it thinks twice. LLMs generate answers in one pass. make a mistake early, the whole thing
GENERATIVE AI
-
Model outputs text and images with visible CoT
By
–
There’s text CoT (visible in the AI Studio UI) but my understanding is it’s from a single model that outputs both text and images
-
Nano Banana: Refreshing Alternative to Corporate AI Branding
By
–
I love that it's called Nano Banana instead of Google Imagine or some other corpo thing
-
AI Generates Customized Weekly Workout Plan with Visual Illustrations
By
–
I first asked it to create a workout plan as a separate text interaction up front – I gave it some of my preferences, goals, etc. and asked for a weekly plan. Then I asked to illustrate the week, and the individual days after.
-
AI Creates Personalized Weekly Workout Plans and Printable Posters
By
–
I asked it to create a personalized weekly workout plan, and then posters that I can print on the wall to remind me what exercises to do each day. Tuesday looks more intense because I asked for "more testosterone" :D.
(sorry I'll stop posting more nano banana pro stuff now) -
Intelligent Canvas: The GUI Evolution for LLM Interaction
By
–
Imo this is along the lines of how talking to an LLM via text is like typing into a DOS Terminal and "GUI hasn't been invented yet" of some of my earlier posts. The GUI is an intelligent canvas.
-
Gemini Nano Banana Pro Solves Exam Questions from Image
By
–
Gemini Nano Banana Pro can solve exam questions *in* the exam page image. With doodles, diagrams, all that. ChatGPT thinks these solutions are all correct except Se_2P_2 should be "diselenium diphosphide" and a spelling mistake (should be "thiocyanic acid" not "thoicyanic") :O
-
Gemini 3 Progress Rivals GPT-4 Breakthrough Moment
By
–
the Gemini 3 moment seems closer to GPT-4 than any other moment in recent times. So much progress released so suddenly that it's overwhelming. Especially NanoBanana's progress.
OpenAI around GPT-4 looked invulnerable , and Google looks invulnerable now with Gemini 3 (+TPUs -
AI Generates Handwritten Math Solutions from Images
By
–
I confirmed that it works. Prompt was the starting image and "Fix this math problem and generate updated image in my handwriting"
-
GPT-5.1 Pro: Advanced AI Colleague with Enhanced Expertise
By
–
GPT-5.1 Pro: “feels like a step toward models that think and communicate more like real colleagues, with better domain expertise, intuition, and judgment, along with increased empathy and communication skill”