Gemini 2.0 Pro vs Deepseek R1 vs Openai o3 mini | Who will win? : Arch AGI Bench

[@tantzer6113]

R1’s answer at @8:46 can be considered as correct, since the question has more than one right answer. Here’s the alternative answer: the color of the boundary of the rectangle with the largest length. The problem with this benchmark is that the questions don’t always have unique solutions. Updated result: tie between o3 mini and DeepSeek.

[@Trinity_TF2]

im excited to see gemini 2.0 pro thinking

[@YJxAI]

yes me too..

[@Trinity_TF2]

btw if gemini 2.0 pro is that good without thinking then it will be good

[@videosclips_]

Thanks 🎉❤

[@YJxAI]

❤️

[@gemini_537]

Gemini 2.0 Pro ❤

[@bodethoms8014]

Gemini has thinking now

[@michaelspoden1694]

Yeah I was wondering why you didn't use that it actually has two thinking models at the moment

[@YJxAI]

there's only flash thinking guess.

[@Family_Guy_12]

okay ❤

[@imqqmi]

Who will win?
The machines of course!

[@abhrodipsingharoy4508]

I think you should scold o3 mini-high for better response

[@YJxAI]

🤣 yeah.

[@darksidedevelopment]

I've been on the fence about Gemini. It always gives me different results than any other model. I've sorta just dismissed it entirely, but now that they have deep think. I may have to revisit Gemini as a potential tool.

[@YJxAI]

I think you are talking about the gemini right not ai studio.

[@darksidedevelopment]

@@YJxAI Yes, I was referring to the Gemini Chat. It has been very hit and miss, more miss than hit :)

[@YJxAI]

​@ yeah it is becoming compelling lately.

[@tvk-ox2my]

what program you are using to generate those promt

[@YJxAI]

I made a common prompt like understand the below question and each time new question comes it is placed in the placeholder. and I can copy it.

[@wilsonbotlero2363]

Isn't there a thinking model from google too in AI studio?

[@YJxAI]

there is did a video on it. Please do check :)

[@hamloji]

Do manim coding challenge