@thsottiaux Being productive in 5 hours chunks per day wasn’t working. I’d rather lock in and destroy my weekly limits in a day and go touch grass for the rest of the week.
@Blindvfx123 I’m going to blow your mind. I did this with Gemini and claude. And made an app called motion monkey and used the app to make this motion graphic. Crazy times we live in man. I am not a motion designer.
Gemini 3 is crazy. First example is Claude Sonnet 4.5, second example Gemini 3 pro. Same initial prompt was used. It took 4 more prompts with Claude to get it working. Only took 2 for Gemini and it was just to change the background to white.#gemini3#googlegemini#googledeepmind
@OfficialLoganK@tulseedoshi@ancadianadragan@melvinjohnsonp@ZacharyGleicher Something I didn't hear you guys touch on was the function calling performance of the 2.5 models. Flash still has a tendancy to say it called a function when it hasn't. Have there been any improvements on this because it is really unreliable
@OfficialLoganK Gemini's function calling abilities, especially 2.5 flash remains the core reason why i can't switch from Open AI's models. Gemini is so good but boy does it suck at function calling. Sadly even GPT-4o-mini does better.
@OfficialLoganK Are you guys shipping Gemini with audio output anytime soon? Or like Gemini TTS? Also updates to Gemini with images. Maybe 2.5 pro with images or 2.5 flash...
These are oddly specific but I have a crazy app idea I want to build