Full Summary
This Tuesday morning, OpenAI and Anthropic are battling it out with new AI model releases, while multiple nations roll out their own specialized AI. Both TechJuice and TheLec.net confirm OpenAI has launched GPT-5.5, with a preview of GPT-5.6 also unveiled. GPT-5.5 excels in coding, computer use, and scientific research, scoring 82.7% on Terminal-Bench 2.0. It can handle multi-step tasks with less human guidance, planning and verifying its own output. However, TheLec.net reports the public release of GPT-5.6, described as OpenAI's highest-performing model to date, is delayed by a U.S. government request. Meanwhile, Anthropic has launched Claude Sonnet 5, as reported by Anthropic, TechCrunch, and Gizmodo. This new model is designed to be highly "agentic," meaning it can make plans, use tools, and run autonomously, performing at a level previously seen only in larger, more expensive models. TechCrunch highlights that Sonnet 5 promises performance close to Opus 4.8 but at a much lower cost, priced at $2 per million input tokens and $10 per million output tokens through August 31st. Technobezz adds that Claude AI models are now available through Microsoft Foundry on Azure, running on NVIDIA's GB300 Blackwell Ultra GPU systems. In parallel, multiple sources including Cybernews and The Portugal News confirm Portugal is launching its own open-source AI model called "Amália," specifically designed for European Portuguese. This model will be deployed in the public sector for education, defense, and healthcare. Similarly, The420.in reports the Institute of Chartered Accountants of India is partnering with Sarvam AI to develop a specialized, secure large language model for chartered accountants, addressing privacy concerns. What nobody expected is China's significant move towards AI self-reliance. Both BeInCrypto and BizzBuzz report that Meituan, a food delivery giant, has debuted LongCat-2.0, an open-source large language model with 1.6 trillion parameters, trained entirely on domestic hardware without NVIDIA chips. This marks the first trillion-parameter model to complete both training and inference on Chinese hardware, outperforming Google's older Gemini 3.1 Pro on some benchmarks. This rapid innovation and national focus on AI ownership means consumers and businesses can expect more specialized, powerful, and potentially more secure AI tools to emerge, impacting everything from professional services to national infrastructure.