Google’s Gemini 3.1 Pro takes aim at GPT-5 and Claude Opus 4.6: What’s new in the AI model?

Google has introduced Gemini 3.1 Pro, an upgrade over its Gemini 3 Pro AI model. The new update offers better reasoning, improved coding performance, and lower operational costs. Google says the model is now rolling out across its consumer apps, developer tools, and enterprise platforms. Also Read: How AI Can Make Your Daily Life Easier

On its official blog, Google shared the details about the new update, where it also compared its performance to other popular AI models in the market including Sonnet 4.6, Opus 4.6, GPT-5.2, and GPT-5.3-Codex. Also Read: Big win for India! OpenAI to open new offices in Mumbai, Bengaluru in 2026

Stronger reasoning and benchmark gains

Google mentions that Gemini 3.1 Pro brings a noticeable improvement in reasoning compared to the earlier Gemini 3 Pro AI model. On the ARC-AGI-2 benchmark, which tests how well models handle new logic problems, the model reportedly scored 77.1 percent. Google says this is more than double the performance of Gemini 3 Pro.

Reports also note that Gemini 3.1 Pro leads Anthropic’s Claude Opus 4.6 on the Artificial Analysis Intelligence Index. Google claims the new model achieved these results while using fewer tokens, which reduces cost for users running large workloads.

The model continues to support a 1 million token context window, along with tool calling, structured outputs, and JSON mode for integration into applications.

Focus on coding and practical tasks

Gemini 3.1 Pro places strong emphasis on code-related reasoning. Google says it has posted improved results on coding benchmarks such as Terminal-Bench Hard and SciCode.

Google also demonstrated practical use cases, including generating website-ready animated SVGs using just text prompts. Since these are created using code instead of video, file sizes remain small while retaining clarity.

Google says hallucination rates have also been reduced compared to the previous preview version, which could improve reliability in real-world tasks.

Where it’s available

Gemini 3.1 Pro is rolling out in preview across several platforms. Consumers can access it through the Gemini app and NotebookLM, with higher usage limits for Google AI Pro and Ultra subscribers.

Developers can try the model through Google AI Studio, Gemini CLI, Android Studio, and Google Antigravity. Enterprise customers can access it via Vertex AI and Gemini Enterprise.

Add Techlusive as a Preferred Source

For now, the model is in preview, and Google says further improvements are expected before general availability.