# News: July 9, 2026

> 11 curated links on AI agents, Python, security, Google Cloud, and engineering, selected by Médéric Hurier.

- Author: [Médéric Hurier (Fmind)](https://www.fmind.dev/)
- Published: 2026-07-09
- URL: https://www.fmind.dev/news/2026-07-09/

- [Introducing Grok 4.5](<https://x.ai/news/grok-4-5>): SpaceXAI has launched Grok 4.5, its most intelligent model ever trained alongside Cursor, specifically optimized to excel at coding, agentic tasks, and software engineering.
- [Introducing GPT-Live](<https://openai.com/index/introducing-gpt-live>): OpenAI has introduced GPT-Live, including GPT-Live-1 and GPT-Live-1 mini, a new generation of low-latency voice models designed for highly lifelike and responsive interactions that now power ChatGPT Voice.
- [Gemma 4 Technical Report](<https://arxiv.org/html/2607.02770v1>): Google has released Gemma 4, a new generation of open-weight, natively multimodal dense and Mixture-of-Experts models ranging from 2.3B to 31B parameters featuring a thinking mode and improved efficiency.
- [Introducing Muse Image and Muse Video](<https://ai.meta.com/blog/introducing-muse-image-muse-video-msl>): Meta has launched Muse Image and previewed Muse Video, its first media generation models developed by Meta Superintelligence Labs that feature agentic tool use, self-refinement, and native audio support.
- [Robostral Navigate: single-camera AI navigation](<https://mistral.ai/news/robostral-navigate>): Mistral AI has introduced Robostral Navigate, an 8B parameters model designed for single-camera AI navigation that achieves 76.6% on R2R-CE benchmarks without requiring depth sensors, LiDAR, or multiple cameras.
- [A global workspace in language models](<https://www.anthropic.com/research/global-workspace>): Anthropic researchers have discovered a "J-space" within Claude's internal neural activations that acts as a global workspace for silent, flexible reasoning and can be monitored or edited to trace a model's hidden thoughts and intents.
- [How to deploy an AI agent to Cloud Run (step by step)](<https://www.youtube.com/watch?v=bjZ2M7ThOBc>): Google Cloud Tech has released a step-by-step video tutorial explaining how developers can build and deploy containerized AI agents to Cloud Run.
- [Facing US export controls, China's DeepSeek plans to make its own chips](<https://arstechnica.com/ai/2026/07/facing-us-export-controls-chinas-deepseek-plans-to-make-its-own-chips>): China's DeepSeek is planning to develop its own in-house data center chips for AI inference to reduce reliance on Nvidia and Huawei, following a year of engineering recruitment and partner discussions.
- [Microsoft Replaces OpenAI, Anthropic With Own AI in Some Apps](<https://www.bloomberg.com/news/articles/2026-07-07/microsoft-replaces-openai-anthropic-with-own-ai-in-some-apps>): Microsoft has begun replacing certain OpenAI and Anthropic AI features in Excel and Outlook with its own in-house MAI models to reduce high API licensing costs, according to Bloomberg.
- [Beijing is looking at curbing overseas access to China's top AI models, sources say](<https://www.reuters.com/world/beijing-is-looking-curbing-overseas-access-chinas-top-ai-models-sources-say-2026-07-07>): Chinese authorities have held meetings with major tech firms, including Alibaba and ByteDance, to discuss restricting international access to China's most advanced proprietary and open-weight AI models.
- [Claude is definitely not conscious...](<https://www.youtube.com/watch?v=KGtko3y2RXQ>): Fireship has released a video breaking down Anthropic's J-space interpretability paper and evaluating whether Claude exhibits signs of machine consciousness.