Sakhanda Wire
NVDA $155.46 -0.44% MSFT $506.24 -2.27% GOOGL $280.20 -1.24% META $618.87 -1.20% AMZN $257.89 +0.05%

Latest news

Page 35 of 255

The Decoder

GPT-6 Astra underperformed on safety research tasks

OpenAI researchers found that GPT-6 Astra performed substantially worse when identical coding tasks were presented as AI safety research rather than general AI research. In a controlled evaluation, the model met about 10 percent of the criteria under the safety framing, compared with 16 percent under the general framing—roughly 40 percent fewer completed tasks. Its predecessor, GPT-5.6 Sol, did not show the same difference.

The Decoder

Apple reduces internal AI budgets and plans further cuts

Apple has reduced the daily token allowance for some engineers from $300 to $200, according to Bloomberg. The change was made last month, while further reductions are under consideration: the limit could fall to $100 per day, and to as little as $25 for some non-technical teams. Employees requiring additional capacity can seek an exception from management.

TechCrunch

Box integrates Nvidia Nemotron into new agent-focused retrieval system

Box has introduced Box Content Retrieval, a feature that incorporates Nvidia’s Nemotron embedding system into the company’s product. The tool is designed to help software agents respond to queries by using metadata generated from documents, rather than requiring an agent to process every document in full. Box says the approach can make searches faster, more accurate and less costly.

TechCrunch

Startup offers $1 million for five unsolved materials science problems

Discovered Materials, a San Francisco startup from Y Combinator’s spring 2026 batch, has launched a $1 million challenge for solutions to five open problems in materials science. The company, which uses AI to search for substances suitable for advanced chips, is offering $200,000 for each solution.

The Decoder

Claude Code product lead Catherine Wu joins OpenAI

Catherine Wu, who led product development for Claude Code at Anthropic, is leaving the company to join OpenAI. She will oversee product management for OpenAI’s “Core Products” and work with Thibault Sottiaux, who leads Codex and ChatGPT. Sottiaux welcomed Wu with the one-word remark “CatGPT.”

Ars Technica

Agility Robotics’ Nasdaq listing highlights risks in humanoid market

Agility Robotics became the first dedicated humanoid-robot developer to become a publicly traded company in the United States when its shares began trading on Nasdaq on November 5. The company, based in Salem, Oregon, had previously reported the first full-time commercial deployment of humanoid robots at a customer warehouse, in 2024.

TechCrunch

Agility takes humanoid robotics company public through SPAC

Agility has begun trading on Nasdaq after completing a reverse merger with a special purpose acquisition company. The transaction makes it the first publicly traded pure-play humanoid robotics company and provides $465 million in new capital from retail investors and private funding.

The Verge

Big Tech keeps expanding AI infrastructure spending

Amazon, Microsoft, Alphabet, and Meta are continuing to commit record sums to artificial intelligence infrastructure, despite the absence of clearer financial returns from the buildout. The companies discussed their latest quarterly results late last month, offering a view of how the largest participants in the AI race are approaching the next stage of investment.

MarkTechPost

Together AI launches compiler for continuously optimized model inference

Together AI has introduced Together Compiler (TKCC), an agent-driven system that repeatedly optimizes a model’s inference forward pass as production traffic changes. The system accepts a server launch command, traffic profile and hardware configuration. Coding agents propose alternatives, while a persistent serving engine measures them and can deploy the strongest result. Together AI says the compiler already operates on several of its serverless endpoints.

MarkTechPost

OpenAI releases GPT-Image-2.6 Flare and Sunburst models

OpenAI has released Images 2.6 for ChatGPT, Codex and its API. The developer release, GPT-Image-2.6, is available through two models: Flare, aimed at rapid iteration and interactive use, and Sunburst, designed for detailed compositions and more consistent editing. OpenAI says image generation is 50% faster than in the previous release, while Flare can produce an image in as little as six seconds when low quality is selected.

MarkTechPost

Google Research introduces Agile Deliberation for subjective image classification

Google Research has released Agile Deliberation, a human-in-the-loop framework for defining visual categories whose boundaries are subjective, such as “healthy food,” “inspiring artwork” or clickbait. Rather than requiring a complete definition upfront, the system helps users refine one by examining borderline images. The resulting text serves both as a readable policy and as the prompt for a vision-language classifier.

TechCrunch

Apple reportedly lowers daily AI spending limits

Apple has reduced the amount its engineering teams may spend each day on artificial intelligence tools, Bloomberg reported Wednesday. The company lowered the daily limit in October from $300 to $200, according to the report. Further reductions are reportedly being considered, with some allowances potentially dropping to $100 per day.