Alibaba's Qwen3.7-Max Autonomously Optimizes Code, OpenAI Launches PowerPoint Plugin

Alibaba's AI model optimizes code for custom chips, OpenAI integrates ChatGPT with PowerPoint, and Deepseek maintains its discounted pricing.

Today's AI news features advancements in autonomous code optimization, productivity enhancements through AI-powered plugins, and cost-effective AI model pricing. Alibaba's Qwen team has released a model capable of long-running autonomous tasks, while OpenAI is streamlining presentation creation with a new ChatGPT plugin. Deepseek's commitment to affordable pricing further democratizes access to powerful AI tools.

🤖 Alibaba's Qwen3.7-Max Achieves Autonomous Code Optimization

Alibaba's Qwen team has launched Qwen3.7-Max, a proprietary AI model designed for long-running autonomous agent tasks. According to *The Decoder*, this model excels in optimizing code for Alibaba's custom chips, running autonomously for 35 hours. Benchmarks show that Qwen3.7-Max matches Claude Opus 4.6 and outperforms Chinese rivals like DeepSeek V4 Pro and Kimi K2.6. The team also demonstrated the model's capabilities by using it to steer a four-legged robot.

📝 OpenAI Integrates ChatGPT with PowerPoint via New Plugin

OpenAI has introduced a new beta plugin that brings ChatGPT directly into PowerPoint, as reported by *The Decoder*. This add-in allows users to create presentations from notes, documents, or images, and also enables editing of existing slides. The plugin is available worldwide across all tiers of ChatGPT. OpenAI recommends saving important presentations before using the plugin.

💰 Deepseek Makes 75% Discount Permanent, Undercutting GPT-5.5 Pricing

Deepseek is making its 75 percent discount on its top model V4-Pro permanent, according to *The Decoder*. This pricing structure sets the cost at $0.435 per million input tokens, making it at least 11.5 times cheaper than GPT-5.5 on input and over 34 times cheaper on output. This move could significantly benefit token-hungry agentic systems by reducing operational costs. Deepseek is reportedly prioritizing AGI research over short-term profits, despite securing billions in funding (*The Decoder*).

👓 Google Demos Prototype AI Glasses with Gemini-Powered Overlays

Google recently demoed prototype Android XR glasses that overlay Gemini-powered translation, navigation, and other information directly into the user's field of view, according to *TechCrunch*. These glasses represent a step towards seamless integration of AI into everyday life, providing real-time assistance and information. The demonstration suggests a future where AI enhances our perception and interaction with the world.

📚 UC Berkeley Law Implements Restrictions on AI Use in Legal Education

UC Berkeley Law will ban AI from nearly all graded work starting in summer 2026, as reported by *The Decoder*. This includes outlining, drafting, and proofreading, with only research use remaining allowed. The school's rationale is that future lawyers must first develop independent thinking skills before effectively using AI.

What this means: These developments highlight the ongoing evolution of AI, from enhancing productivity tools to optimizing hardware and shaping educational approaches. The integration of AI into PowerPoint and the development of AI-powered glasses demonstrate the increasing accessibility and practicality of AI in various domains. Deepseek's pricing strategy and Berkeley Law's AI restrictions reflect a growing awareness of both the opportunities and potential challenges presented by AI.

The AI landscape continues to evolve, with a focus on both practical applications and the ethical considerations surrounding its use.