OpenAI Launches Beta, Claude Arrives on Google, AI Gap at 3%

Between beta APIs, office integrations, and infrastructure configurations, the week has delivered concrete milestones for AI adoption. The gap between the United States and China is quantified at 3%, and players are refining their security measures.

Technological Competition and Security: 3% Gap and Verified Access

The gap in artificial intelligence between the United States and China stands at 3%, with DeepSeek V4.1 Flash leading the benchmarks for agentic coding. China is facing challenges in attracting foreign AI researchers, which could impact its technological development. Anthropic is expanding verified access to its cyber capabilities, enhancing security and transparency in the use of its technologies. Additionally, AICR v1.0 offers an open, stable, and verifiable GPU cluster configuration, facilitating access to computing resources.

Ready-to-Use Tools: Decision APIs and Claude Compatibility

The decision API offered by OpenAI is now available in an open beta version to the public. This interface aims to simplify the automation of decision-making in various situations. Claude, an AI model, is becoming compatible with Google Docs, Sheets, and Slides, increasing its utility in workplace environments. Meanwhile, Hark has launched an AI agent a year before the release of its first devices, illustrating rapid advancements in technological development.

Models and Debates: Innovations, Mathematics, and Emergency Responses

Mistral Large 4 is presented as an AI model that promises significant advancements. Nano Banana 2.1 brings notable improvements and aims to optimize performance in specific applications. The sharing of AI progress in mathematics continues to grow, highlighting the capabilities of algorithms in this field. The discourse on cybersecurity risks is currently deemed inadequate, calling for a reevaluation of approaches and strategies. Building faster emergency correction systems is considered essential to respond to the increasing threats in the technological landscape. Finally, the question of which AI is in production and who owns it remains an important topic of debate.