AI models

The EU has empowered its executive branch to significantly scrutinize AI models, restrict market access, and impose substantial fines on providers. This move intensifies oversight of major U.S. AI companies like Anthropic and OpenAI, impacting their ability to deploy advanced AI systems in the bloc. These new powers, part of the EU AI Act, aim to assert technological sovereignty and manage the risks associated with increasingly powerful AI.

OpenAI has significantly reduced prices for its GPT-5.6 Terra and Luna models by 20% and 80% respectively. This strategic move aims to make advanced AI more accessible and cost-effective for enterprises, addressing the growing demand for clear ROI and countering competition from more economical solutions, especially from China. The priciest model, Sol, remains unchanged.

Microsoft CEO Satya Nadella questioned Anthropic’s strict limitations on its Fable AI model, deeming them illogical and hindering its utility as a creation tool. This critique arises amidst Microsoft’s significant investment in Anthropic and its reliance on their models for Copilot. Nadella emphasized the need for cost-effective AI and warned against a market dominated by a few token-rich companies. This also reflects Microsoft’s move to develop more in-house AI solutions.

Alphabet shares fell 4% following reports of a delay in its Gemini AI model release, particularly for advanced coding. This setback highlights intense competition in AI development, with rivals like OpenAI and Meta showcasing superior code generation capabilities. Google is prioritizing performance enhancement, especially in complex code generation, for Gemini 3.5 Pro. The delay underscores the challenges of scaling advanced AI and maintaining a competitive edge in a rapidly evolving market.

Chinese AI models are gaining traction in the U.S. due to competitive performance and significantly lower costs. These models are closing the gap with U.S. rivals, with some U.S. companies seeing their AI model usage share from China exceed 40%. This trend is driven by escalating token prices for U.S. models, making cost-effective Chinese alternatives increasingly attractive for businesses seeking to scale AI initiatives.

The US Department of Commerce has lifted export controls on Anthropic’s Claude Fable 5 and Mythos 5 AI models, resolving a dispute with the Trump administration. This decision allows Anthropic to redeploy these advanced models after a temporary suspension due to national security concerns. The move comes amid rapid advancements in Chinese AI and aims to maintain US leadership in the field.

OpenAI has launched three new advanced AI models: GPT-5.6 Sol, Terra, and Luna. Access is initially restricted to trusted partners following a U.S. government request. OpenAI aims for broader public release soon, emphasizing collaboration with the government on pre-release assessments. GPT-5.6 Sol is their most sophisticated model, with notable advancements in coding, biology, and cybersecurity. This move highlights the evolving balance between AI innovation and governmental oversight.

OpenAI is developing its own custom “Jalapeño” chip with Broadcom to address the immense costs of scaling AI models like ChatGPT. This move aims to improve efficiency and reduce reliance on third-party hardware, mirroring Apple’s vertical integration strategy. The chip is optimized for LLM inference, minimizing data movement. OpenAI’s accelerated chip design process leverages its own AI models, with initial deployment planned for late 2026.

Amazon is intensifying its generative AI efforts, aiming to challenge industry leaders like OpenAI and Anthropic with frontier models within a year. The company is focusing on building a strong foundation in data, architecture, and infrastructure, alongside developing proprietary models like Nova2. A key differentiator is Amazon’s integrated chip strategy, with custom silicon like Trainium and Graviton designed to optimize AI performance, positioning them as a significant player in the AI race.

The U.S. government has mandated Anthropic cease access to its advanced AI models, Fable 5 and Mythos 5, citing national security concerns. The directive, issued without prior indication of escalating threats, forced Anthropic to disable customer access. The company views the government’s main concern as a “potential narrow jailbreak” and disputes the necessity of recalling commercial models for such an exploit, warning of industry-wide implications. Anthropic is in discussions to resolve the issue.