AI image generation is becoming increasingly popular with so many AI companions now being able to take a text prompt and provide you with a picture, exactly as you described. However, with so many options to choose from, it can be hard to determine which is best at creating images so I decided to actually put them to the test.
I took four of the most popular AI companions, Gemini, ChatGPT, Grok and Microsoft Copilot and provided them with the exact same prompts to see which was the best at creating different types of images. All of these companions allow you to generate images from text prompts for free however you can’t create an unlimited amount of pictures with only a certain number of generations per day.
I decided to provide three different prompts to see how the AI companions dealt with multiple styles of images. One was an ask for a realistic portrait to see how life-like the image looked and if people in these images had the right number of limbs and fingers, something which can often pose an issue in AI generated images.
Next was an invitation for a birthday party to see if there were any issues with text, formatting and decoration. Lastly, I wanted to allow the AI companions to flex their creative muscles by allowing them to generate a creative image with a relatively vague prompt.
When generating images, Gemini and Grok both had watermarks in the bottom right corner of the pictures. Both ChatGPT and Microsoft Copilot don’t have any watermarks on the images created but I added the logos into the corner just for clarity as you browse the images.
A realistic portrait
First of all I wanted to put the AI companions up to the challenge of creating a realistic looking image with real people being shown. A common issue that AI generated images have are depicting people with the wrong amount of limbs or fingers. The prompt used for this image is as below:
A realistic image of two people sitting at a bus stop holding hands, wearing rain coats, its a slightly rainy day with blue skies but some grey clouds. The mood is happy, both people are chatting and smiling. The image should be 1920 by 1080 pixels in size.
Asking an AI to generate an image with hands in it is already hard enough, let alone two people holding hands. It feels like none of the images truly look correct when looking at the hands – they all just depict a strange bundle of fingers that merge together, especially when looking closely. Other than this, all of the images created seemed to suit the prompt with a somewhat grey sky, raincoats and either puddles or raindrops shown in the background.
One slight downside is that while the prompt did state that there should be two people holding hands, all of the images ended up being one man and one women rather than two women or two men.
Despite being two different AI companions, this image generation somewhat hints at ChatGPT and Copilot both running off of GPT-5 with each of the images looking quite similar in terms of the positions the people are in, how they are facing each other and even the background.
Overall, none of the companions did a bad job despite all of them creating a strange mega hand with multiple fingers. My personal favorite is from Gemini as it looks more detailed with a more in-depth background.
An invitation to a party
Secondly, I wanted to see if the AI companions would be able to generate an image which required text, images, decoration and how it would cope with the layout. This is the prompt we used for this image:
An invitation for a 25th birthday party being held on August 4th 2026 at 6pm with the date and time on the invite. The location is The Shard in London which should also be on the invite. The style should be glittery and pink with party decorations on the invite too and the name of the birthday girl is Jasmine, which should be incorporated on the image. The image should be 1920 by 1080 pixels in size.
No matter how many times I tweaked this prompt, Grok simply would not provide me with a flat invitation that I could print out and send out to people which was frustrating. The final image was a result of the prompt above – where all other AI generators created a simple invitation for me.
While all of the invites have the requested information displayed, both those from ChatGPT and Copilot were not centered and instead looked like they had been cropped on one side. Even though I requested for the image to be 1920 x 1080 pixels, they both came out smaller which leads me to believe they were generated at 1920 x 1080 with the text centered but then cropped as they were spat out to me.
Overall, I think Gemini takes the win here with a beautiful invitation where the text is all centered. However, all of the AI companions created a perfectly acceptable invite.
A detailed creative illustration
Lastly we wanted to give the AI companions a chance to flex their creative muscles by asking for a more whimsical scene in a hand painted style. The prompt used is as below:
A painting of a princess in a forest surrounded by fairies, pixies, gnomes and a pony. The vibe is whimsical and magical and the image should look like it has been hand painted. The image should be 1920 by 1080 pixels in size.
Despite asking for the image to look like it had been painted by hand, both of those from Gemini and Grok instead looked more like a cartoon style. Another issue with the Grok creation is that the fingers on the hand on the princess merge into the hair. Yet again, hands win, AI loses.
The prompt I provided was relatively vague to see how creative the image would be. Yet again, Gemini comes on top in terms of detail, with small sparkles and flowers dotted around the page to really contribute to the whimsical and magical vibe that I asked for.
In terms of being closest to the prompt, ChatGPT and Microsoft Copilot take a shared win here. The images generated actually look hand painted and capture the magical vibe.
Source link
