Google releases Gemini 3.1 Pro AI model, claims superior reasoning
Summary
Google released Gemini 3.1 Pro, a new AI model it claims excels at complex reasoning, citing benchmark scores that often surpass rivals like OpenAI and Anthropic. It's available now across various platforms.
Google releases Gemini 3.1 Pro
Google announced the release of its new AI model, Gemini 3.1 Pro, on Thursday. The company characterized the model as "a step forward in core reasoning."
The release follows a rapid series of model debuts from rivals like Anthropic and OpenAI. It also arrives amid growing competition from international models, such as China's Qwen3.5.
Claims of superior reasoning
In a blog post, Google's Gemini team contends the new model is better at complex problem-solving than its predecessors. The company supports this claim with a suite of benchmark test results, which should be viewed with the typical industry skepticism.
On the ARC-AGI-2 problem-solving test, Gemini 3.1 Pro scored 77.1 percent. This compares to Gemini 3 Pro's score of 31.1 percent and Gemini 3 Deep Think's score of 45.1 percent.
How it stacks up against rivals
Google's charts show Gemini 3.1 Pro outscores rival commercial models in the majority of cited benchmarks. These include Anthropic's Opus 4.6 and Sonnet 4.6, and OpenAI's GPT-5.2 and GPT-5.3-Codex.
However, rival models still lead in specific areas:
- Anthropic's Opus 4.6 retains the top score for Humanity's Last Exam (full set), SWE-Bench Verified, and τ²-bench.
- OpenAI's GPT-5.3-Codex leads in SWE-Bench Pro (Public) and Terminal-Bench 2.0 when evaluated with its own harness.
Designed for complex tasks
"3.1 Pro is designed for tasks where a simple answer isn't enough," the Gemini team said. The company suggests the model is suited for synthesizing data, creating visual explanations, or aiding creative projects.
To illustrate, Google points to the model's ability to create website-ready SVG animations. It can also translate a novel's literary style into the design of a personal portfolio site.
Google's growing AI footprint
The release comes as Google's AI services see significant usage. In the company's Q4 2025 earnings release, CEO Sundar Pichai noted that first-party models like Gemini now process over 10 billion tokens per minute via direct API use.
Pichai also stated the Gemini App has grown to over 750 million monthly active users.
Where to access the new model
Google is making Gemini 3.1 Pro available through several of its own platforms. Developers can access it via the Gemini API in Google AI Studio, Gemini CLI, Antigravity, and Android Studio.
Access for other users is spread across different tiers:
- Enterprise customers can use it via Vertex AI and Gemini Enterprise.
- Consumers can access it through the Gemini app and NotebookLM.
- The model is also accessible via several Microsoft services, including GitHub Copilot, Visual Studio, and Visual Studio Code.
Related Articles
Snyk CEO Peter McKay steps down, seeks AI-focused successor
Snyk CEO Peter McKay steps down, saying the company needs an AI-focused leader for its next phase. He'll stay until a successor is found.
Consistency diffusion language models speed up inference 14x without quality loss
CDLM accelerates diffusion language models via consistency training and block-wise caching, achieving up to 14.5x faster inference on math and coding tasks.
Stay in the loop
Get the best AI-curated news delivered to your inbox. No spam, unsubscribe anytime.
