SoL-Pi cuts coding agents' token usage by up to 49 percent with little change in performance by optimizing the control layer between the model and its environment. A research agent tested 152 approaches across more than 3,000 runs to develop the system, though the gains were smaller on other benchmarks.

The article Nvidia's SoL-Pi system cuts coding agent token usage nearly in half by optimizing the harness appeared first on The Decoder.

Tencent's Gander processes speech, images, and text while handling tasks in the background. A "cerebellum" keeps the conversation going, while a swappable "brain" searches files, writes code, or tackles other complex work. Users can interrupt or change the task mid-conversation. In benchmarks, Gander interrupted users in just 8 percent of cases, less often than its rivals, but trailed on task accuracy.

The article Tencent's Gander aims to keep talking while it works in the background appeared first on The Decoder.

Runway wants to stream AI video as users prompt it, rather than make them wait for finished clips. The approach builds on GWM-1, its world model that generates video frame by frame. Beyond creative tools, Runway sees uses in robotics and autonomous driving.

The article Runway wants to turn AI video generation into a live stream you control in real time appeared first on The Decoder.

Microsoft and the University of Illinois built StudentSim to replicate individual students from limited data and give AI tutors fast, low-cost feedback. In tests covering 60 students across chess, English, and math, it outperformed GPT-5.4. A chess tutor trained with StudentSim also earned the highest expert ratings among three versions tested.

The article Simulated students that make realistic mistakes help AI tutors learn faster appeared first on The Decoder.

Unity has released official plugins for Claude Code and OpenAI's Codex.

The article Unity launches official plugins for Claude Code and OpenAI Codex to stop AI agents from using outdated tutorials appeared first on The Decoder.

Google and Deepmind's Dream-RSI lets AI agents "dream" through past search runs to test new strategies without costly recalculations. In tests, it matched or beat existing results, cutting iterations by a factor of up to 2.43. Only the search strategy adapts, while the underlying AI model stays unchanged.

The article Google Deepmind's Dream-RSI helps AI agents improve by “dreaming” about past attempts appeared first on The Decoder.

The AllSpark team has released Iris-mini and Iris-pro, two open-source search agents built on Qwen models that lead benchmarks among open-weight models in their size classes. According to the paper, the training data and models also improved performance on tasks they were never trained for, including general tool use and office work.

The article Iris-mini and Iris-pro are the strongest open-weight search agents in their class appeared first on The Decoder.

A law professor spent two years testing how an AI ban, unguided AI use, and structured training affect student performance. The group without AI finished last both years. "I was wrong," the researcher writes, who had assumed that AI without guidance would do more harm than good.

The article Two-year university study finds banning AI from classrooms leaves students worse off appeared first on The Decoder.

Reasoning steps like calculation, formula retrieval, and deduction are clearly separable in a model's internal states, especially in the middle layers. That matters for AI safety, because models process more than their visible chain of thought reveals.

The article AI models' written reasoning steps correspond to distinct internal patterns, a new study finds appeared first on The Decoder.

Google Research has released TimesFM-3, a forecasting model that analyzes time series alongside related data and known future events like sales promotions or weather forecasts. Instead of predicting the future step by step, the 330-million-parameter model fills in all future time points in a single pass, which cuts compute time and reduces compounding errors.

The article Google's new AI model predicts the future from sales data, weather, and discount schedules appeared first on The Decoder.

Deepseek releases V4.1-Flash, a multimodal model with 552 billion parameters that cuts KV cache memory to a quarter of its predecessor. On the DeepSWE coding benchmark, it narrowly beats Opus 5 and GPT-5.6 Sol, even though only 16 billion parameters are active per token. The model ships under the MIT license and targets much cheaper AI agents.

The article New Deepseek model V4.1-Flash cuts memory needs for AI agents appeared first on The Decoder.

Meta unveils Muse, an AI agent that books travel, handles purchases, and sends emails through WhatsApp, complete with a payment feature that runs through Stripe's Link. That puts Meta ahead of OpenAI, which stopped its direct checkout feature in ChatGPT. A separate security agent called Sentinel monitors every action before it reaches the internet.

The article Muse can shop, write emails, and negotiate prices for users, all through WhatsApp appeared first on The Decoder.

Nvidia and Palantir want to run supply chains with AI.

The article Nvidia and Palantir team up to run supply chains with AI, starting with Nvidia's own million-part operation appeared first on The Decoder.

Suno has unveiled a new AI music model generation, v6, in three versions, built together with Warner Music Group, BMG, and Believe. All older models are being shut down. Songs can now be partially changed through text commands or generated multimodally from text, audio, and images. The company won't say which catalogs went into training, while Universal and Sony keep suing.

The article Suno launches v6 music models built with Warner, BMG, and Believe appeared first on The Decoder.

Qualcomm is designing custom chips for AWS across multiple product generations, with a focus on AI inference.

The article AWS is using Qualcomm for AI inference while Qualcomm uses AWS Bedrock to design the chips appeared first on The Decoder.

Hugging Face launched "ML Intern," an AI assistant built into its chatbot that lets users run machine learning experiments without any ML expertise.

The article Hugging Face's new ML Intern lets anyone run machine learning experiments through a simple chat appeared first on The Decoder.

Three years after launch, Mistral AI has closed a 3 billion euro Series D round, pushing its valuation past 21 billion euros.

The article Mistral AI raises 3 billion euros in Europe's largest-ever tech funding round despite lagging behind rivals appeared first on The Decoder.

Alibaba's research arm has released Qwen-Drive 1.0, an AI model that handles environmental perception, traffic Q&A, and route planning in one system. The researchers show that text-image models don't automatically understand three-dimensional space. Spatial awareness has to be trained on purpose. The goal is a single model that runs both the cockpit and the driving system.

The article Qwen-Drive 1.0 tells you why it brakes, just don't expect the explanation to match the maneuver appeared first on The Decoder.

Google Research and DeepMind are releasing WeatherNext 3, a weather model that skips traditional physics simulations and learns directly from real-time satellite data. It produces hourly forecasts at up to five-kilometer resolution, five times more detailed than its predecessor. Google says regions in Africa, Latin America, and the Asia-Pacific that have lacked accurate forecasts should see the biggest gains.

The article Google's WeatherNext 3 ditches physics simulations and learns weather directly from live satellite data appeared first on The Decoder.

Meta's Superintelligence Labs have released Muse Voice Transcribe, a real-time transcription model that processes speech in 80-millisecond chunks, tells speakers apart, and detects sentence boundaries. According to Artificial Analysis, it delivers the most accurate streaming transcription at the lowest price in the market. Meta sees the model as a building block for personal AI agents that listen in on real conversations through devices like its camera glasses.

The article Meta's new real-time audio model is the foundation for AI assistants that never stop listening appeared first on The Decoder.