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

Older stories

Hugging Face

Microduck Walking Policies Move from Isaac Lab Simulation to Real Robots

Pollen Robotics and Hugging Face have released a workflow for training the open-source Microduck bipedal robot in simulation and transferring the resulting walking policy to physical hardware. The tutorial uses NVIDIA Isaac Lab 3.0 and Newton to model servo response, friction, delays and gearbox backlash, factors that can affect performance outside simulation.

The Decoder

Google Cloud adds OpenAI models alongside Gemini and Claude

Google Cloud has begun offering three OpenAI models on the same platform as its Gemini and Claude systems, making it the only major cloud provider described in the report as carrying all three model families. The new options are GPT-6 Astra, GPT-6.1 Sol and GPT-6 Luna, which are entering a private preview for selected enterprise customers. Applications for access are open.

Hugging Face

VocabX cuts measured inference cost for an essay grader by 77%

A vocabulary-specialization method called VocabX reduced the measured inference cost of a Portuguese essay grader by 77.0%, according to results from a paper that is still under review. The system uses Qwen3-4B, but its original tokenizer was designed for many languages and domains, leaving it with a 151,669-entry vocabulary and fragmented output for the grader’s repeated JSON structures.

TechCrunch

Blacksmith reaches $1 billion valuation with Sequoia-led round

Blacksmith has announced a $66 million Series C investment led by Sequoia, putting the software platform’s valuation at $1 billion. The company develops tools for building and testing software, including code produced by AI. Its system is designed to support continuous integration, which combines code changes with automated building, testing, merging and validation.

The Decoder

Demis Hassabis narrows his AGI forecast to two or three years

Demis Hassabis, founder of Google DeepMind and a Nobel laureate in chemistry, now expects artificial general intelligence to emerge within two to three years. He defines AGI as a system capable of matching the complete range of cognitive abilities of the human brain. Earlier this year, his estimate was three to five years.

Hugging Face

TorchNeuron profiling compares three RMSNorm implementations on AWS Trainium

A profiling guide published on October 21, 2026, examines how to analyze inference performance with TorchNeuron, the native open source PyTorch backend for AWS Trainium. The first part assumes no previous profiling experience and describes how to configure torch.profiler with the Neuron backend, capture a trace and interpret device and CPU timings.

Ars Technica

OpenEvidence reaches $15 billion valuation as doctors adopt its AI search tool

OpenEvidence, a Miami-based startup founded by serial entrepreneur Daniel Nadler in 2022, has reached a reported valuation of $15 billion as its medical AI service gains widespread use among US physicians. The company says many of the country’s 1 million doctors submitted more than 42 million prompts through the platform last month.

The Verge

Gemini brings conversational navigation to Apple CarPlay

Google is extending its Gemini assistant to Apple CarPlay, nearly a year after introducing the feature on Android Auto. The conversational navigation function is available to drivers in the United States, while Google says a wider global rollout is also underway. The company has not specified which additional regions are included so far.

Hugging Face

IBM updates Granite Speech models for accuracy, speed and speaker attribution

IBM has released three updates to its Granite Speech collection: the new granite-speech-5.0-2b and revised versions of granite-speech-5.0-470m-turboctc and granite-speech-4.1-2b-plus. The models target multilingual transcription and translation, fast edge or on-device processing, and recordings involving multiple speakers.

The Decoder

Mercury 3 claims faster performance than GPT-6 Luna and Claude Haiku 5.5

AI startup Inception has released Mercury 3, a diffusion-based language model that creates a rough version of an answer and refines it in several passes. The company says this approach allows the model to generate more than 1,000 tokens per second on Nvidia GPUs, four to eight times faster than GPT-6 Luna and Claude Haiku 5.5 in its comparisons.

The Decoder

Mistral and Nvidia partner on AI for industrial design

Mistral and Nvidia are joining forces to apply artificial intelligence to engineering simulations and industrial product development. According to Mistral’s announcement, the companies will combine the Mistral Large 4 language model and Mistral’s physics AI with Nvidia’s CUDA-X and Warp software libraries, as well as the Omniverse platform for 3D simulations.

TechCrunch

AI Makes Employee Evaluation Harder for Many Managers

A Nielsen survey of 1,700 directors and senior managers in seven countries found that more than half now consider employee evaluation more difficult after their organizations adopted AI tools. The survey was commissioned by work-management company Monday.com. Nearly half of managers said they spend more time reviewing and guiding employees’ work, much of which reaches them in a polished form that does not clearly reveal the contribution of the employee or the AI system.