Gemini AI: A Powerful Tool with Notable Image Generation Limitations
Table of Contents
Google’s AI chatbot, Gemini, presents a compelling offering in the increasingly crowded artificial intelligence landscape. Unlike many competitors, Gemini doesn’t operate on a standalone subscription model; instead, it’s integrated into the Google One AI Pro plan – previously known as Gemini Advanced and the Google One AI Premium subscription. While the bundled price point may appear substantial, it becomes more accessible for users already invested in the Google ecosystem and utilizing services like Google Drive.
Gemini distinguishes itself through speed, video creation capabilities, and robust tools for developers, but lags behind competitors in text and image generation, according to recent testing. The service delivers a varied quality of results, sometimes exceeding expectations and other times falling short.
Image Generation Struggles Highlight Key Weakness
The most significant drawback identified during testing centers around Gemini’s inability to reliably generate images. One analyst noted the system consistently failed to produce an illustration depicting a residential electrical installation featuring solar panels, electric vehicle chargers, and a home battery system. While no other tested service flawlessly executed this complex request, several generated reasonably useful visuals. Gemini, however, simply responded with “Sure, here is your illustration” – without providing an image.
This issue extended to attempts to create a network chart, with Gemini refusing to generate any output despite repeated prompts and the use of the image generation button.
Comic Book Images and Video Stand Out
Despite broader image generation challenges, Gemini excels in specific areas. The AI produced one of the best results when tasked with creating a comic book image. While text within speech bubbles presented a minor issue – a common problem across all tested platforms and easily corrected – the overall quality was impressive. A request for a “realistic rat eating a Snickers” resulted in a drawn rat consuming something labeled “Snicks,” demonstrating a slight inaccuracy in brand recognition.
Gemini truly shines in video creation. Utilizing the Google VEO 3 model, the AI successfully generated a video of a squirrel discovering chewing gum and blowing a bubble, accurately following the prompt’s instructions. In comparison, ChatGPT and Perplexity AI could render the squirrel, but not the bubble.
However, ChatGPT’s Sora offers a distinct advantage with its integrated editing environment, allowing for fine-tuning and iterative development. For professional video work, Sora may be a more practical choice, depending on individual workflow preferences. For users prioritizing video, exploring specialized tools like Adobe Firefly is also recommended.
A Boon for Developers
Google provides exceptional tools for developers seeking AI assistance. Beyond the core Gemini chatbot, the inclusion of Jules, an AI agent capable of independently tackling programming tasks, is a significant benefit. Jules integrates with Github, automatically addressing issues labeled with “Jules.” Furthermore, Google’s Gemini Code Assist, an extension for VS Code, offers faster performance than standalone AI chatbots, though it isn’t as seamlessly integrated as Microsoft’s Github Copilot.
Specifications and Pricing
Gemini was tested in September 2025 and is accessible via gemini.google.com. Pricing is currently set at SEK 255 per month or SEK 2,600 annually. The service supports uploads of text files, Office documents, PDFs, and pictures, and is available on both Android and iOS. Gemini utilizes the Gemini 2.5 Pro, VEO (Video), Imagen 3, and Gemini 2.5 Flash (Pictures) models, alongside the programming assistants Jules and Gemini Code Assist.
Overall, Gemini receives a grade of 3.5 out of 5, reflecting its strengths in video and developer tools, offset by its limitations in broader image generation and occasional formatting issues.
