OpenAI quietly updated its image generation model. The improved image quality, the removal of the pee filter, and better physical understanding are impressive, but its multilingual performance and reference consistency still fall short of beating Nano Banana Pro.
@arrakis_ai
-
Why GPT 5.1 Underperforms on ARC-AGI-2 Benchmark
By
–
ARC-AGI-2 is being solved incredibly fast, but I still don’t understand why GPT 5.1 scores so low even though it’s the newest model
-
Google Nano Banana Hackathon Feature Submission Experience
By
–
I actually submitted a very similar feature to the Google Nano Banana hackathon about four months ago, so it’s fun to see something like this appear now — though I can’t help feeling a bit disappointed that my project didn’t get selected back then. pic.twitter.com/yyJzCpwBOW
— CHOI (@arrakis_ai) 21 novembre 2025I actually submitted a very similar feature to the Google Nano Banana hackathon about four months ago, so it’s fun to see something like this appear now — though I can’t help feeling a bit disappointed that my project didn’t get selected back then.
-
Google Releases Gemini 3 Pro Image Preview Free
By
–
https://
zenmux.ai/google/gemini-
3-pro-image-preview-free
… -
Gemini 3 Removes Video Upload Feature from AI Studio
By
–
I’m disappointed that video upload has disappeared from AI Studio. Filming the feature I wanted and uploading it to do vibe-coding was an amazing experience, so I’m wondering why Gemini 3 doesn’t support it even though its vision capabilities have improved. I really hope the
-
Gemini 3 Deep Think Achieves 45.1% on AGI Benchmark
By
–
WTF.. Gemini 3 Deep Think hit 41 percent on HLE and reached 45.1 percent on the ARG-AGI-2 benchmark.
-
iOS Version and TestFlight Beta Availability Plans
By
–
Any plans for an iOS version? I’d love to join a TestFlight build
-
Improved AI Responses with Visible Thinking Phase
By
–
I’m wondering if I accidentally flipped some feature flag or something (or maybe I’m in a beta test), because it definitely feels different. The results are way better, but having the “thinking” phase appear first makes it feel slower, even if the actual response time hasn’t
-
GPT-5 Shows Improved Routing Accuracy and Overall Performance
By
–
i feels like GPT-5’s routing accuracy and overall performance have improved a lot. Is it just me?
