Google has just released its new Gemini 3 Pro AI, claiming it can outperform almost all other advanced AI models. We decided to test it against OpenAI’s latest ChatGPT 5.1 Thinking model. Both models were given extra “thinking time” to perform at their best. Here’s how they compared.
In logical reasoning tests, both Gemini 3 Pro and ChatGPT 5.1 Thinking solved a tricky puzzle correctly. They identified John as the bald man and agreed there was no need to text an apology.
When it came to a more complex riddle about sandwiches, Gemini 3 Pro quickly gave the correct answer, while ChatGPT 5.1 Thinking took over four minutes and gave the wrong answer. This showed Gemini 3 Pro’s stronger reasoning skills.
For website creation, both models performed well. Gemini 3 Pro quickly researched and generated HTML, CSS, and JavaScript code, while ChatGPT 5.1 Thinking took longer but included more detailed content. Both results were usable and modern.
In visual reasoning tests, like generating an SVG of a pelican riding a bicycle, Gemini 3 Pro produced a more realistic and accurate image. ChatGPT 5.1 Thinking’s version had errors in posture and leg positioning.
Gemini 3 Pro also excelled at generating a realistic 3D spinning Rubik’s Cube with proper shadows. ChatGPT 5.1 Thinking’s code did not run correctly and only showed a blank background.
In a clinical reasoning test, both models correctly recommended Spironolactone as the suitable diuretic for a patient with hypokalemia, showing both are strong in medical reasoning.
Overall, Gemini 3 Pro impressed with faster reasoning, better coding, and stronger visual outputs. Unlike previous Gemini models, it is concise, accurate, and highly usable. At this point, Google’s Gemini 3 Pro appears to have overtaken ChatGPT 5.1 Thinking as the leading AI model in performance and versatility.
