GPT-5.6 🚀, Muse Spark 1.1 ✨, ChatGPT Work 💼

OpenAI released the GPT-5.6 series on July 10, 2026, comprising three models: Sol, Terra, and Luna. Among them, Sol is the flagship model, demonstrating leading intelligence and efficiency in coding, cybersecurity, and scientific fields, and is the first model to win the ARC-AGI-3 public game. According to TLDR AI, Sol can correctly understand unfamiliar scenarios and use the game's own vocabulary, marking a breakthrough in environmental adaptability.

The specific positioning of Terra and Luna has not been detailed yet, but the overall series emphasizes better performance with fewer tokens and lower costs. GPT-5.6 also supports multi-agent parallel processing and improved design judgment capabilities, further enhancing applications in cybersecurity and scientific research.

Alongside the new model release, OpenAI shut down its Atlas browser, integrating agentic browsing capabilities into the ChatGPT desktop app and Chrome extension. This decision is part of OpenAI's broader effort to reduce side projects and focus on core products.

Additionally, OpenAI launched ChatGPT Work, a collaborative workspace powered by GPT-5.6 that gathers context from team tools and performs actions across files and desktop applications, aiming to transform scattered project materials into complete documents, spreadsheets, and presentations.

OpenAI's second-in-command executive, Fidji Simo, plans to step down from her full-time executive role due to a worsening neuroimmune disease, transitioning to a part-time advisor. Her product and business responsibilities will be divided among President Greg Brockman, CFO Sarah Friar, and Chief Strategy Officer Jason Kwon. This is the latest in a series of recent leadership changes at OpenAI.

OpenAI faces serious allegations in a copyright lawsuit. According to TLDR AI, news organizations accuse OpenAI of repeatedly lying for years to hide evidence of infringement. OpenAI privacy engineer Vincent Monaco revealed that the company misled the court for two years, claiming it could not search ChatGPT logs, even though such searches had been conducted before the lawsuit began. This discovery could lead the court to impose severe sanctions on OpenAI.

Meta released Muse Spark 1.1, improving tool use, coding, computer interaction, and multimodal reasoning, and opened the Meta Model API in public preview.

Anthropic appointed former Federal Reserve Chair Ben Bernanke to its Long-Term Benefit Trust to help address the economic impact of AI. Additionally, Anthropic invited public input on AI-related concerns and conducted surveys and focus groups to understand societal expectations and fears.

Furthermore, Elon Musk publicly praised Anthropic's Mythos/Fable model and promised not to 'cut off' relations with Anthropic.