Sakhanda Wire
NVDA $152.56 -2.29% MSFT $505.79 -2.36% GOOGL $279.70 -1.42% META $618.14 -1.32% AMZN $256.03 -0.67%

Latest news

Page 8 of 253

The Decoder

Anthropic CEO warns of uncertainty ahead of planned IPO

Anthropic CEO Dario Amodei has described the company’s spending on computing capacity as its “deepest business dilemma” shortly before its planned IPO on Nov. 19. Speaking with JPMorgan CEO Jamie Dimon at an investor event in New York, Amodei said capacity must be secured one to two years before the company can know whether future revenue will cover the cost. He added that no one yet knows how rapid AI advances will translate into economic value.

The Decoder

Gemini 4 Carbon tops MathArena at a much higher cost

Google’s Gemini 4 Carbon has taken first place in the MathArena mathematics benchmark, becoming the first non-OpenAI model in some time to lead the ranking. It displaced OpenAI’s GPT-6.1 Sol, while Anthropic’s Claude-Fable-5.2 moved into second place. OpenAI’s models are not publicly available yet, although they have reportedly handled demanding problems such as the Navier-Stokes equations.

MarkTechPost

Prototype brings weight-space programming to DSPy

Cmpnd has released dspy-wsp, a prototype optimizer that adds weight-space programming to DSPy. Its WeightSpace component lets a frontier language model inspect and directly modify the weights of a smaller model, then evaluate each change. The process uses no gradients, training loop or fine-tuning dataset. The package is MIT-licensed, can be installed with pip, and requires DSPy 3.5.

The Decoder

Anthropic urges users to update prompts for Claude Opus 5.5

Anthropic says users should abandon several prompting practices developed for older model generations when working with Claude Opus 5.5. The model is now the default on the Pro, Max, Ultra, and Team paid plans. The company says phrases such as “think hard” and “think carefully” are unnecessary because Opus 5.5 reasons automatically and adjusts its depth to the task and selected effort level. In Anthropic’s tests, removing those phrases made responses begin faster without a measurable quality loss.

TechCrunch

Asian governments assess Nvidia after Jensen Huang’s departure

Governments across Asia are responding differently to Jensen Huang’s departure from Nvidia. The company’s board opened an independent investigation on Sunday, November 8, following reports that Huang had been targeted by a Chinese spy operation. Huang also acknowledged an affair with Alice Huang, a former executive at Singapore-based chip company Megaspeed.

The Verge

Helen Toner says AI safety cannot depend on trusted CEOs

Helen Toner, executive director of Georgetown University’s Center for Security and Emerging Technology, says AI safety should not rely on the judgment or trustworthiness of any single company leader. Toner previously served on OpenAI’s board and was one of four directors who voted to remove Sam Altman. Altman later returned as CEO, Toner resigned, and OpenAI replaced its board and restructured. The episode exposed the limits of a nonprofit oversight mechanism intended to keep safety ahead of profit and growth.

The Verge

Jensen Huang’s public image shifts from star founder to controversy

Nvidia CEO Jensen Huang received a highly theatrical welcome at CES 2025 in Las Vegas on January 6. He entered a 14,000-seat arena wearing a black leather jacket and raised one hand as the audience applauded and cheered. Influencers and fans filled the venue, which was illuminated by a blue glow.

The Decoder

OpenAI redistributes robotics researchers across its other teams

OpenAI is reportedly dissolving its standalone robotics team for the second time, but the move is not described as a withdrawal from robotics. Researchers are being transferred to the main research group led by Chief Research Officer Mark Chen, while applied robotics and hardware staff are joining Greg Brockman’s Applied organization. The decision came so quickly that a booth booked for the Conference on Robot Learning in Austin was left unused, Fortune reported.

The Decoder

Coxon says Anthropic can retrain Mythos with far less compute

Former OpenAI and Anthropic researcher Jacob Coxon says a slide presented during Anthropic’s IPO roadshow was a major reason he left the company after roughly four months. The graphic tracks the computing resources required to train models over time. Its compressed, logarithmic scale makes tenfold and hundredfold increases appear relatively similar, yet the curve still rises sharply. Coxon interprets this as evidence that algorithmic efficiency is improving at an accelerating pace.

MarkTechPost

Stanford Researchers Release Memorilla Memory System for Frozen LLMs

Stanford researchers have released Memorilla, an open-source memory module designed to help frozen large language models use large document collections. The system compresses a collection into 16 learned memory tokens, which are then added to a question for Qwen3-8B. The decoder and Qwen3-Embedding-4B encoder remain frozen, while the 194,123,776-parameter memory module is trained.

The Decoder

Runway releases open robotics model Praxis-1

Runway has released Praxis-1, a robotics model with open weights and source code under the Apache 2.0 license, which permits commercial use. The system combines camera images, text instructions and a robot’s position data to predict changes in its surroundings and produce corresponding movements. It updates its control as new camera frames arrive.

MarkTechPost

Vercel Labs launches beta UI SDK for AI coding agents

Vercel Labs has released Vercel UI SDK, an open-source beta library containing more than 50 interface behaviors. It is headless: instead of finished components, it provides primitives for keyboard navigation, focus, selection, popups, dismissal, drag and drop, and related ARIA requirements. Coding agents can combine them into accessible interfaces for React, Vue, Svelte, Solid and Web Components. The library is distributed across six npm packages, with 51 component modules in the repository.