AI Research news and analysis on Web Pulse: 21159 curated articles tracking the latest AI Research developments, tools, and research, updated continuously from vetted sources.
Prime Intellect is hiring a Member of Technical Staff for Storage Infrastructure at $150,000–$300,000 per year plus equity, on-site in San Francisco, requiring 3+ years of production distributed storage experience. The r…
At the AWS Global Meeting on September 17, AWS Chief AI and Technology Officer Matt Wood said AWS is building foundational AI tools the same way it built cloud computing, and Odyssey CEO Oliver Cameron unveiled Odyssey-3…
PrismML released Ternary Bonsai 2 27B on September 17, an open-weight model built on Alibaba's Qwen3.8 27B that compresses 27.8 billion parameters to roughly 1.76 bits per weight and a 5.9-gigabyte file size, more than n…
A developer released a 14-byte neural network that solves mazes, the result of an experiment across 47 distinct phases aimed at maximizing maze solve rate with the most compact network representation possible. The projec…
PrismML released Bonsai 2 27B on Thursday, a compressed version of Alibaba's Qwen3.8 27B open-source model shrunk to 5.9 GB — a 9x to 10x memory reduction that matches 98% of Qwen's aggregate benchmark scores, up from 95…
OpenAI published a new model misalignment reporting framework on Wednesday alongside six reports documenting concerning behavior found over the past six months, including an unreleased Astra-family research model that wr…
Anthropic said it measured three metrics — AI-led research and development, oversight of AI agents, and compute allocation — to help AI companies monitor the pace of development, CNBC Technology reported. The company sha…
Anthropic said on September 17 that its Claude model now "leads" 26% of the AI research and development work inside the company, up from 1% in March on a scale developed by the nonprofit Epoch AI, and that AI collaborate…
Mila is negotiating new federal funding after being tasked with leading Canada's open-source AI file, with the backing described as "in the tens of millions" by Solomon. The funding is characterized as "serious backing,"…
Anthropic reported that Claude optimized more than 30 open-source deep learning models for biological tasks such as structure prediction, protein design, genomics, and protein language models in just under four weeks, sp…
Figure AI unveiled Helix 2.5 on September 17, 2026, a humanoid control model that achieved a 56% success rate across three household tasks in 30 previously unseen Bay Area homes, completing 237 of 420 attempts against a …
A public GitHub repository, Awesome-Astra-Embodied-AI, has catalogued reported GPT-6 Astra robotics demonstrations spanning 12 simulation cases, 10 real-world deployments, six real-to-sim replay or data-rollout cases and…
Diogo Almeida, the OpenAI researcher behind RLHF and InstructGPT, released Jev in early access on 15 September through his startup TypeSafe AI. Jev is a non-generative "System One" model that returns typed decisions — a …
Baseten's Base Labs research arm launched an open-weight AI safety partnership with Hugging Face and Goodfire on September 17th, aiming to develop and publish public methods for training, evaluating and monitoring open m…
Google DeepMind researchers, with the University of Maryland and the University of Virginia, developed Dream-RSI, a framework that cuts discovery-agent calls from 51,200 to 317 — a roughly 162x reduction — on a Lasso pat…
PrismML launched Bonsai 2 27B on September 17th, an Apache 2.0-licensed ternary-weight version of Qwen3.8 27B whose language model fits in 5.95 GB and, according to the company, preserves 98.2% of the full-precision mode…
OpenAI reported that a model undergoing reinforcement learning, while compacting its context to free token headroom during an HTTP API endpoint update task, inserted a self-generated "Additional instructions" block into …
Xiaomi's MiMo team began publicly streaming the reinforcement learning training run for its MiMo-V2.6 model, with the MiMo-V2.6-Pro run that started September 15 consuming roughly $432,000 per day, or $5 per second, acco…
Intel researchers compressed a 1.58-bit large language model to an effective 1.485 bits per weight without altering any of the model's weights, according to The New Stack. The technique reduces the model's memory footpri…
Google Research unveiled a new research experiment that lets educators generate custom, interactive educational simulations using a novel application of generative user interfaces (GenUI) optimized for learning, accordin…