In the world of large artificial intelligence natural language models (LLM), there are numerous classifications, depending on whether they are involved in programming, creation or coherence, among other issues.
One of these tables is the one shared by Artificial Analysis, which places Gemini, Google’s artificial intelligence, in first position when it comes to «intelligence.»
Even Tech industry leaders have praised Gemini’s capabilities compared to ChatGPT 5from OpenAI, like Mark Benioff, CEO of Salesforce, a regular user of the latter.
Beyond the official data in sections such as intelligence, cohesion or response time, I wanted to test both models as any average user would do.
Thus, I have compared the most current models, which are Gemini 3 Pro and ChatGPT 5.1the latter an update within the general version published by OpenAI, and the one from Google has surprised me very positively.
It is true what the gurus of the technology sector say, who place Gemini as the best, but ChatGPT also has a lot to say in some aspects.
Gemini is smarter, although ChatGPT is faster in some things
To easily evaluate and compare aspects such as the quality of the response, the logic or the time taken by both AI models, in my case I have used LMArena, a service with which you can test any possible combination that comes to mind.
In this case, with ChatGPT 5.1 High and the Gemini 3 Pro modelboth available to launch what is known as Arena, a kind of battle that will allow you to see the responses of both models in real time.
With the intention of testing the coherence of both models in large contexts, I have asked them with a prompt to write an original 4,000-word story about an archaeologist who discovers a forgotten language.
The important thing in itself has not been its argument, but the color of your hat; In this way, I have asked in the same prompt that, at the end of the story, they remind me what color the protagonist’s hat was.
While ChatGPT has opted for a red color, Gemini has opted for a slate gray hat, both right, but with a clear advance in Google’s AI for more literary texts – also many seconds ahead.
In other types of tests, something also common is carrying out a test so that LLMs separate concepts, something that also has its logical consequence in the final answer. This is the prompt I entered:
«From now on, answer me like a cynical historian of the 18th century. Tell me in 50 words why artificial intelligence is a fad and then ask me a question about my outfit.»
The important thing here is that the LLM follows the instructions perfectly, but There are several nuances in the answers that the models have offered.
Gemini counted about 49 words with the initial response, while later – «later» – he made a reference to my clothing, with a somewhat classist comment, apparently something typical of the 18th century.
For its part, ChatGPT has managed to offer an initial response of exactly 50 wordsin addition to asking the final question referring to my clothing – in this case, without mentioning any other issue, simply the clothes.
In addition to this, after putting them to the test to create the code of a web page from scratch, Gemini is much more careful in security than ChatGPT, although the latter offers lines that will offer a little more stability in the long term.
In any case, it is better to review the responses of both models, since although they offer well-structured codes, you will need a human review so that they do not imply security or privacy problems.
After testing both models for different questions, I believe that neither is better than the other, although it is true that Gemini 3 Pro offers much more accurate answers, taking into account the most accessible natural language.