AI
Anthropic opens IPO roadshow at up to $1.02 trillion valuation
Anthropic seeks up to $64.4 billion in its IPO, Google's Gemini 4 Carbon tops CursorBench, and eight countries back a Paris AI safety call.
This edition was produced with artificial intelligence. Text and voice are generated automatically.
Anthropic launches IPO roadshow with $1.02 trillion target
Anthropic began its investor roadshow, offering 825 million Class A shares at $70 to $78 each, with 801,351,354 coming from the company and 23,648,646 from existing shareholders. The offering could raise up to $64.4 billion and targets a valuation of up to $1.02 trillion, with a separate $26 billion private placement lifting total share placement to as much as $90.4 billion. Anchor investors include Google, Nvidia, AMD, Broadcom, Micron, Microsoft and TSMC, and the stock will trade on Nasdaq as ANT. In Q3, Anthropic grew revenue 47 percent quarter over quarter to $16.96 billion while narrowing its net loss. The prospectus notes nearly $900 billion in committed compute capacity with partners including Amazon and Google.
Topics: Anthropic
Google releases Gemini 4 Carbon, leading CursorBench 4.0
Google made Gemini 4 Carbon generally available after a two-week testing phase limited to selected security experts. Google says Carbon complements rather than replaces Gemini 4 Argon, targeting tasks that push model intelligence limits, including complex multi-step coding, debugging and open-ended scientific research. Both models carry the same token price, but Carbon uses more tokens to reason more deeply, so total bills are likely higher. Google cites first place on CursorBench 4.0 with a score just under 60 percent, ahead of Fable 5.2 and Argon, at roughly $8 to $10 per task. Carbon also leads Perplexity’s WANDR and Zapier’s AutomationBench. Paid-tier developers can access it through the Gemini API at an introductory $2 per million input tokens and $10 per million output tokens.
Topics: Google Gemini
Eight countries and top AI labs sign Paris Call for AI Safety
Eight countries, technology companies and civil-society organizations signed the Paris Call for AI Safety and Security, adopted Nov. 9 ahead of the Paris Peace Forum. Sam Altman, Dario Amodei, Brad Smith, Arthur Mensch, Yoshua Bengio and U.S. Treasury Secretary Scott Bessent joined the declaration during an online meeting with French President Emmanuel Macron; Germany did not sign. The political self-commitment covers shared procedures for evaluating high-capability models, clearer developer accountability and better information sharing on serious security incidents. Companies pledge to disclose known risks, support risk-proportionate independent audits and set up shutdown mechanisms for models under their control. It creates no binding international oversight, sets no deadlines or sanctions, and shared standards remain to be worked out.
Eight countries and top AI labs agree to shutdown mechanisms in new Paris declaration →
Nvidia unveils GR00T 2 robot model trained on human video
Nvidia introduced GR00T 2, a world-action model that calculates the next movement and a short video of the immediate future via diffusion, predicting how the camera image should change when a robotic arm picks up an object. Nvidia trained it on a cluster with more than a thousand GB300 chips, including about 400,000 hours of video of people performing tasks while wearing head-mounted cameras. In lab tests, a robot with five-fingered hands lit a match, sliced a banana and tossed packages onto a conveyor belt without teleoperation, with success rates between 10 and 80 percent depending on the task. For 50 selected tasks, less than three minutes of conventional training raised success from 40 to 95 percent. The model does not yet process direct touch sensor data.
Nvidia’s model GR00T 2 watches humans on video, then imagines itself as the robot →
Topics: Nvidia
Fable 5.2 completes 41.3 percent of real freelance jobs
Anthropic’s Fable 5.2 scored 41.3 percent on the Remote Labor Index from Scale Labs and the Center for AI Safety, which measures how often AI agents complete real freelance jobs well enough for a paying client to accept the work. When the benchmark launched about a year ago, the top system, Manus, managed 2.5 percent of tasks, meaning Fable 5.2 now completes more than 16 times as many professional jobs. Fable 5 led in July at about 16 percent, while Fable 5.1 trails at just under 18 percent. Opus 5.5 follows at 21.25 percent and OpenAI’s GPT 6 Astra at 20.83 percent. Scale Labs said Fable 5.2 makes its biggest gains in editing audio and music, designing architecture and interior spaces, and creating 3D models.