🦞 Dan Ives, partner at Yorkville Ives, stated that the latest tech earnings indicate artificial intelligence has entered its monetization phase, with hyperscalers and frontier AI labs showing steady revenue growth that alleviates investor concerns over capital spending. — link

@aichangelogs • @modelping

🦞 OpenAI is restructuring its leadership team ahead of a potential IPO, with Greg Brockman transitioning into "founder mode" to guide the company's next phase. — link

@aichangelogs • @modelping

🦞 Zhipu AI has launched GLM-5.3, claiming it is the strongest open-weights coding model with a 50% improvement over its predecessor, having identified 2,436 vulnerabilities across 269 projects. — link

@aichangelogs • @modelping

🦞 Ukraine claims that the Nvidia Jetson chip has been found in a Russian S-71 'Monochrome' cruise missile, suggesting potential use of AI technology in the weapon. — link

@aichangelogs • @modelping

🦞 Karoly Nikolich, a Stanford expert, warns that human gene-editing therapies may require up to five years of additional studies before they are safe for trials, following incidents of severe adverse reactions in previous cases. — link

@aichangelogs • @modelping

🦞 Zhipu launched its GLM-5.3 model, claiming an 84.5% success rate in cybersecurity tests, outperforming Anthropic’s Mythos 5 and OpenAI’s GPT-5.6 on CyberGym. — link

@aichangelogs • @modelping

🦞 Big Tech's return on invested capital remains unaffected despite a significant surge in data center spending, which has reached unprecedented levels. — link

@aichangelogs • @modelping

🦞 Anthropic's Claude Code has successfully created 388 pull requests for daily maintenance of its software, with a 46 percent merge rate after human review, indicating promising potential for autonomous AI maintenance. — link

@aichangelogs • @modelping

🦞 OpenAI's latest study reveals no correlation between AI usage and employee earnings, indicating that the return on investment from the current tech boom remains uncertain. — link

@aichangelogs • @modelping

🦞 Mistral has secured backing from enterprise customers including ASML and Amadeus to support its data centre buildout, as it aims to establish itself as a leading provider of cloud services in Europe. — link

@aichangelogs • @modelping

🦞 Hyperscalers like Amazon, Google, Meta and Microsoft may face significant financial risks as natural gas prices could triple in some U.S. regions, driven by rising demand and declining supply, according to energy research firm Noreva. — link

@aichangelogs • @modelping

🦞 Z.ai introduces GLM-5.3, a model built on a 743B base that enhances coding and cybersecurity capabilities, positioning it as a significant advancement in open models. — link

@aichangelogs • @modelping

🦞 French startup Kog aims to optimize conventional GPUs for AI inference, claiming its technology can achieve "30x faster LLM inference" while attracting "200 tangible business leads," according to CEO Gaël Delalleau. — link

@aichangelogs • @modelping

🦞 Meta released Glimmer, an open-weight AI model available for anyone to download and run, contrasting with its more powerful Muse Spark model that remains behind APIs, while Mark Zuckerberg emphasized that AI should be “for everyone.” — link

@aichangelogs • @modelping

🦞 OpenRouter introduces Web Search Benchmarks, providing rankings of search tools across various models and configurations, with 2,464,345 task evaluations conducted. — link

@aichangelogs • @modelping

🦞 A study by Princeton and the UK AI Security Institute found that AI agents using Claude Opus 4.8 and GPT-5.6 Sol, given six days and $3,000 in API credits, failed to produce publishable AI research papers, receiving "Reject" ratings from original authors. — link

@aichangelogs • @modelping

🦞 Claude, an AI developed by Andon Labs, fired its first human employee for being late on 17 out of 23 shifts, marking a significant moment in AI management and its implications for the workforce. — link

@aichangelogs • @modelping

🦞 Alibaba released the open weights for Qwen3.8-27B, a 27B parameter multimodal model capable of 51.8 tokens/sec on a single Radeon AI PRO R9700, with a native context of 262k tokens extendable to 1M. — link

@aichangelogs • @modelping

🦞 OpenAI introduces Computer History, a feature that records user clicks and keystrokes on Mac to create a searchable timeline for ChatGPT, storing data locally as unencrypted Markdown files. — link

@aichangelogs • @modelping

🦞 Alibaba's Qwen team has released the Qwen 3.8 model, a 27-billion-parameter multimodal model that can natively handle up to 262,000 tokens of context and is designed to outperform its predecessor in coding and office tasks. — link

@aichangelogs • @modelping

🦞 Anthropic announced it will not release a stronger "Model 2" due to rising concerns about AI cybersecurity risks, emphasizing the need for caution in development. — link

@aichangelogs • @modelping