AI News Daily
AI developments selected from public sources
- 6
- major stories
- 13
- sources
- 4
- first-party stories
Lead story
Anthropic S-1 discloses rapid valuation rise and composition of $80.7 billion net loss
Anthropic's S-1 shows that Google and Amazon agreed in March and April to buy shares at $259 per share, while weeks later new investors paid $589. Amazon's early loans can convert into shares at $2.09 to $5.05; as the valuation rose, these fixed-price rights appreciated, and accounting rules recorded them as an Anthropic loss totaling $72.4 billion plus $9.4 billion, accounting for nearly all of its $80.7 billion net loss in the first half of 2026; the company had no cash outflow, and the business itself lost only $700 million. Broadcom, which leases AI chips to Anthropic, has also agreed to finance part of the rent.
Model releases / updates
Baidu open-sources Unlimited OCR 2 document parsing model
Baidu has open-sourced Unlimited OCR 2, which uses a new SR-SWA mechanism to compress the attention interaction between text and vision into a fixed-size working reference, thereby parsing documents of almost any length within a limited context window. The company says per-step decoding latency and memory usage remain nearly constant as documents grow longer, and it achieves SOTA performance on MPDocBench and MarathonDocBench, which cover documents ranging from dozens to hundreds of pages.
Product release/Updated
Tencent Hunyuan open-sources real-time Chinese-English simultaneous interpretation model Hy-SimulMT
Tencent Hunyuan has open-sourced the real-time Chinese-English simultaneous interpretation model Hy-SimulMT, with about 3 seconds of latency and support for real-time voice cloning. The company says it outperforms most open-source models and commercial APIs, and the model has been released on Hugging Face, ModelScope, and GitHub.
Apple Creator Studio update: Final Cut Pro built-in MotionVFX, Logic…
Apple has released an update for Apple Creator Studio: Final Cut Pro for Mac now includes built-in MotionVFX effects and templates, and timeline images can be edited directly in Pixelmator Pro. Logic Pro for Mac adds an intelligent assistant (beta) powered by a combination of Apple's in-house models and OpenAI, and Stem Splitter can separate drum sounds; Pixelmator Pro introduces Generate Fill for modifying selected images with natural language. Keynote, Pages, and Numbers add an assistant (beta) that understands document context and support MCP, allowing external AI agents to read content, perform edits, and automate operations.
Industry news
Nvidia plans to invest $3 billion for a stake in Meta, deepening AI assistant Muse cooperation
Nvidia and Meta announced a strategic partnership, with Nvidia planning to invest $3 billion in Meta's equity offering, and the two sides will deepen cooperation in the AI assistant Muse and large-model fields. Meta will expand deployment of Nvidia's Vera Rubin architecture, Vera CPU, Spectrum-X Ethernet networking technology, and BlueField-4 devices, and discuss secure deployment and Sentry technology based on the OpenShell reference architecture. Jensen Huang and Mark Zuckerberg both said AI agents represent a new computing paradigm, and the two sides will optimize infrastructure end to end to lower total cost of ownership.
Apple releases new 14-inch MacBook Pro with M6 chip
Apple releases new 14-inch MacBook Pro with M6 chip, featuring a 12-core CPU, 12-core GPU, and dual 16-core Neural Engine, with battery life up to 24 hours. Compared with M5, graphics performance is up to 30% faster, and on-device AI workloads such as LLM prompt processing are up to 25% faster; upgrading from M1 brings up to 11x faster graphics and up to 13x faster AI performance. The new machine comes standard with 16GB unified memory, configurable up to 32GB, supports Wi-Fi 7 and Bluetooth 6, opens pre-orders on October 28, and goes on sale November 6.