|
☁️ Cloudflare launches open-source Microsoft 365 rival
LINK
|
- Cloudflare has open-sourced Cloudflare OS, an AI-native office suite where employees ask an agent to generate documents, dashboards, spreadsheets, workflows, and custom apps instead of using a fixed set of tools like Microsoft 365 or Google Workspace.
- The platform builds custom apps called Gadgets, each with its own UI, server logic, API, and data, and routes models through Cloudflare AI Gateway so admins can pick providers, set budgets, and reserve pricier models for harder tasks.
- Security centers on Gatekeepers that grant agents zero access by default and track where data travels, and Cloudflare reports 4,000+ apps built internally in 30 days, though it labels the release early access with acknowledged rough edges.
|
💻 Meta launches AI coding agent Muse Code
LINK
|
- Meta has launched Muse Code, its first AI coding agent, in beta, pushing into the software-engineering assistant market dominated by OpenAI and Anthropic.
- The agent handles tasks like writing code and validating results, powered by new models from Meta Superintelligence Labs under Chief AI Officer Alexandr Wang.
- Meta also shipped Muse Spark 1.2, a coding-focused update to its primary model, though its models have lagged OpenAI's ChatGPT and Anthropic's Claude on coding.
|
🧠 Hassabis exits DeepMind CEO role for AGI
LINK
|
- Demis Hassabis is stepping back from his Google DeepMind CEO role to become Chair of GDM and Chief Scientist of Alphabet, freeing him to focus full-time on shaping the path to AGI.
- Koray Kavukcuoglu, previously GDM's CTO and the founder of its deep learning team behind WaveNet and DQN, takes over as SVP overseeing Gemini development, Frontier AI research, and the Gemini app and developer teams.
- Hassabis stays CEO of drug-discovery spin-off Isomorphic Labs and continues advising GDM on AGI strategy, noting the team's progress on new models including Gemini 4.
|
💸 DeepSeek warns of big price hike
LINK
|
- DeepSeek has warned developers that it will "significantly" raise prices across its API services soon, though it disclosed neither the size of the increase nor an effective date, according to a Bloomberg-reported notice.
- DeepSeek-V4-Flash currently runs $0.14 per million input tokens and $0.28 per million output, rates that pressured ByteDance and Tencent to cut their own after May's permanent V4 discount, marking DeepSeek's second pricing change in under a month.
- The notice landed a week after DeepSeek-V4-Flash-0731 shipped, a 284B-parameter lightweight model, though the company hasn't said whether the hike applies equally to V4-Flash, V4-Pro, cache-hit pricing, or reasoning models.
|
|
