Google's Project Suncatcher prototype satellite successfully completed on-orbit commissioning, generating its first AI response using an onboard Gemma 4 model and data-center-class TPU. This milestone validates the basic feasibility of running AI compute chips in space, paving the way for future thermal/radiation stress tests and the planned twin-satellite interconnection in 2027.
Open-source LLM
开源大模型
Large language models with public weights allowing free community modification and use
Curated Intelligence
Nvidia has acquired open-source hub Hugging Face and acqui-hired teams from code-generation startups Poolside and Essential AI. This strategic consolidation aims to assemble top-tier talent and infrastructure to train leading US-based open-source models, with Nvidia currently eyeing Reflection AI as a potential next target.
Nvidia is negotiating the acquisition of open-source model startup Reflection AI, aiming to integrate frontier algorithmic R&D directly into its hardware ecosystem. If completed, this move signals a strategic shift from pure infrastructure provision to full-stack vertical integration of 'chips + models'. While specific valuation and terms remain undisclosed, the deal could reshape competitive dynamics between open-source communities and commercial closed-source models.
German search engine Ecosia has terminated its partnership with Mistral due to concerns over quality lag, server instability, and environmental compliance, switching instead to Chinese open-source models like Qwen, GLM, and Kimi. This strategic pivot halved Ecosia's AI service costs while improving performance, highlighting the growing competitiveness of Chinese open-source models in European commercial applications.
LinkedIn engineers published a practical retrospective on implementing Agentic Reinforcement Learning (RL) for the GPT-OSS model. The article details optimizing multi-step decision-making through direct environment interaction, moving beyond single-turn responses or static datasets. This work offers reusable technical pathways and lessons learned for training agentic capabilities in open-source LLMs.
Hugging Face published the first in a series of blogs reflecting on the 'DeepSeek Moment' since the release of R-1 in January 2025. The article analyzes how this event drove strategic shifts in China's open-source community, an explosion of new open models, and ecosystem reshaping.