The next level of agentic AI requires high-level reasoning, Google released Gemini 3.1 Pro on Tuesday to deliver it.
Google said that Gemini 3.1 Pro represents the next step in the core logic. The model is better at complex problem-solving and for tasks for which a simple answer is not sufficient. Along with coding tasks, the model can generate website-ready animated scalable vector graphics (SVG) directly from a text prompt. SVG are images created with instructions instead of pixels.
The release of Gemini 3.1 Pro follows the introduction on February 17 by generic AI vendor and Google rival Anthropic. sonnet 4.6A model that focuses on better coding and computer usage skills. This also comes a week after Google upgraded Gemini think deeplyIts thinking model for solving challenging research problems.
This release also highlights a trend in the AI market in which model providers are pushing towards the next phase agentic ai And saying that better logic will help enterprises get there. Since reasoning is a key part of agentic AI, a model with better reasoning skills is a better AI system than one with lower levels of reasoning. reasoning ability.
“In principle, it’s true that if you want to do complex tasks, you have to be able to reason better,” said Chirag Shah, a professor at the University of Washington’s School of Information.
However, better reasoning is not the only building block needed for complex tasks, Shah said. Furthermore, it’s not clear what Google means by “complex”.
“Just saying, ‘Oh this model can do complex tasks,’ you have to take it seriously because it depends on how you define complex tasks,” he said.
But Gemini 3.1 Pro also demonstrates other strong capabilities, such as the ability to ingest, understand and consume data from APIs, and the ability to code simulations using integrated tools, said William McKeon-White, an analyst at Gartner.
“It’s good, continued progress,” McKeon-White said. “However, it was not something I would say was a fundamental gamechanger.”
Apart from claiming better reasoning skills, Google, by introducing updated iterations of its original base model, Gemini, is also showing how it is building models that meet multiple enterprise needs. Although Google is not yet known as the preferred AI vendor for a specific task due to the anthropic coding it is known for, the tech giant aims to show enterprises that it is still adept at those tasks.
“They want to be a one-stop shop for all your modeling needs,” Shah said. He said Google is trying to build a portfolio that will attract enterprises that may care about coding, search and informational functions. While enterprises may have the option to choose Anthropic, OpenAI, Google Or other vendors will settle on only one ecosystem to buy the majority. Google is striving to provide well-crafted models that suit different enterprise needs.
