OpenAI Proposes Superintelligence "New Deal" as Anthropic Revenue Hits $30 Billion
AI Models and Releases
Google DeepMind Launches Gemma 4 with Multilingual and MTP Capabilities: Google DeepMind has officially released Gemma 4, a new model family that shows significant performance gains in European languages like Danish, Dutch, and Italian. Researchers also discovered the model includes Multi-Token Prediction (MTP) architecture, which could enable much faster generation speeds if fully utilized.
- What it took to launch Google DeepMind's Gemma 4
- Gemma 4 is a huge improvement in many European languages, including Danish, Dutch, French and Italian
- Turns out Gemma 4 had MTP (multi token prediction) all along
ACE-Step 1.5 XL Model Series Released: The ACE-Step 1.5 XL family has launched, offering three distinct versions: Turbo, Base, and SFT. These models are designed to provide developers with enhanced capabilities and flexible options for high-performance AI projects.
Minimax 2.7 Update Imminent: Preparations are underway for the release of Minimax 2.7, with ongoing infrastructure adaptation work currently being finalized. The update is highly anticipated by the community for expected performance improvements.
AI Policy and Society
OpenAI Proposes "New Deal" for the Superintelligence Transition: Sam Altman has published an industrial blueprint for the "Intelligence Age," warning that superintelligence is imminent and requires a massive overhaul of the American social contract. The proposal includes creating "Public Wealth Funds," adopting 4-day workweeks, taxing automated labor, and establishing AI access as a fundamental right.
- Axios: Sam Altman States Superintelligence Is So Close That America Needs A New Social Contract On The Scale Of The New Deal During The Great Depression
- OpenAI just dropped their blueprint for the Superintelligence Transition: "Public Wealth Funds", 4-Day Workweeks
- https://openai.com/index/industrial-policy-for-the-intelligence-age
- https://cdn.openai.com/pdf/561e7512-253e-424b-9734-ef4098440601/Industrial%20Policy%20for%20the%20Intelligence%20Age.pdf
- https://www.axios.com/2026/04/06/behind-the-curtain-sams-superintelligence-new-deal
- https://youtu.be/B21KxGs8zDI
Business and Industry News
Anthropic Annual Revenue Surges to $30 Billion: Driven by explosive demand for Claude and high-level enterprise adoption, Anthropic’s revenue has reached $30 billion, up from $9 billion late last year. This rapid scaling is supported by strategic compute partnerships with Google and Broadcom.
AI Giants Unite to Protect IP in China: OpenAI, Anthropic, and Google have formed a rare alliance to combat the unauthorized copying of their AI models in China. The collaboration focuses on protecting intellectual property and maintaining a competitive edge against unauthorized model replication.
Developer and Enterprise Implementation
Mistral Integration in Sovereign Fleets and Vision Optimization: Mistral AI models are being utilized in sovereign multi-agent fleets to ensure EU regulatory compliance and autonomous operations. Separately, new research suggests that simple image rotation correction significantly boosts the performance of Mistral-based vision models during preprocessing.