Google Unveils Gemini 3.1 Pro with 77.1% Benchmark Score, Faces Trade Secret Indictment
Alphabet introduced Gemini 3.1 Pro, which achieved 77.1% on the ARC-AGI-2 benchmark, more than doubling the previous version’s reasoning performance and reaching over 740 million monthly users. The company was hit with a 14-count federal indictment alleging former engineers stole over 300 Tensor processor trade-secret files, carrying potential 20-year sentences.
1. Gemini 3.1 Pro Upgrade and Reach
Alphabet launched Gemini 3.1 Pro, highlighting a 77.1% score on the ARC-AGI-2 benchmark—more than double the previous model’s performance. The update is available through the Gemini API, Google AI Studio, Vertex AI, Gemini Enterprise, the Gemini app and NotebookLM, and builds on a user base exceeding 740 million monthly active users.
2. 14-Count Tensor Trade Secrets Indictment
Federal prosecutors unsealed a 14-count indictment claiming two former engineers and an associate stole over 300 files containing Tensor processor trade secrets. Defendants face at least 20-year sentences if convicted, after Google’s internal security flagged the unauthorized downloads and alerted law enforcement.
3. Market and Legal Implications
The Gemini 3.1 Pro upgrade could drive stronger cloud revenue growth and enterprise adoption in the AI race, while the legal action introduces reputational and operational risks around in-house chip development. Investors may weigh the boost to AI credentials against potential stock volatility from the ongoing legal proceedings.