Google is pushing the boundaries of artificial intelligence with the recent announcement of Gemini 3.1 Pro, its latest large language model. The update, arriving less than three months after its previous major release, underscores the rapid pace of innovation in the field.
Breaking with its typical mid-year update schedule, Google announced Gemini 3.1 Pro on February 19, 2026. The company positions the new model as particularly adept at complex and agentic tasks, but the release also serves to maintain Google’s competitive edge in the rapidly evolving AI landscape.
À lire aussi :
Gemini : notre guide pour comprendre l’IA de Google
According to Google’s own testing, Gemini 3.1 Pro outperforms its predecessor, Gemini 3 Pro, released in November 2025. This latest model aims to solidify Google’s position against other conversational AI agents, as noted by Numerama.
From Videos to Websites and Apps
Google states that Gemini 3.1 Pro is specifically designed to deliver “sharp reasoning for the most complex problems.” While details on the model’s capabilities remain somewhat limited, the company explains that it can assist with concrete applications, such as providing clear visual explanations of complicated topics, synthesizing data into comprehensive overviews, and bringing creative projects to life.
To illustrate its capabilities, Google showcased a video demonstrating Gemini 3.1 Pro generating vector images from text prompts. While the comparison to the previous model is favorable, it’s possible the examples were carefully selected to highlight the update’s strengths. The model can also be used to create sophisticated websites and applications, provided users formulate effective prompts.
Across reasoning tests, scientific knowledge, and development tasks, the new version of Gemini marginally outperforms the competition, achieving slight percentage point advantages. This move allows Google to differentiate itself from Claude Opus 4.6 and GPT-5.3-Codex, recently announced models from Anthropic and OpenAI.
What is OpenAI Planning?
Google’s lead may be short-lived, as rumors suggest a new ChatGPT 5.3 could arrive soon. This potential release is reportedly a response to the “red alert” issued by Sam Altman several weeks ago.
Pour aller plus loin
OpenAI : 100 milliards, de la pub et une croissance « folle », la contre-attaque de Sam Altman
For those interested in testing the new features, the free version of Gemini offers variable access to the model, with fewer restrictions applied to AI Pro and AI Ultra subscriptions (priced at €22 and €275 per month, respectively).
