Articles in
New research reveals AI agents excel at ML engineering but lack deep domain insight in scientific discovery, as shown by the Agent4Sci benchmark study.
A human-AI discovery loop has enabled the construction of irreducible polynomials with Mahler measures below Boyd's limit point for every even degree up to 2000, surpassing the prior limit of degree 180. The team released a verified dataset and introduced the Generalized Boyd algorithm.
AlphaDiana reveals why adding tools and memory to AI agents often fails. Discover how harness-aware evaluation exposes hidden accuracy trade-offs in LLMs.
Boost coding agents reliability with SWE-bench insights. Learn how teaching location and edit diversity improves efficiency and reduces compute costs.
Proxy Confidence audits black-box LLM agents using surrogate models. Achieve higher accuracy than self-reporting with this open-source auditing method.
Discover why Praxa AI agents consume 37% more tokens than baselines to ensure reliability through deterministic verification and strict execution gates.
Discover ZGCM-1, a fully open-source 7B LLM rivaling Qwen3. Learn how its hybrid attention and agentic search boost efficiency and performance.
Explore the AlphaGenome Atlas: Google DeepMind's one-petabyte map of 9 billion DNA variants. Free access for academic research via a user-friendly web portal.
Anthropic has introduced the Model Hardware Standard (MHS), a framework designed to enable AI agents to interact with physical machinery like robots and lab equipment. The system uses standardized interfaces and safety tags to manage real-world constraints, with initial testing underway among select research partners.
Nvidia and Stripe lead a $26B surge in open-weight AI acquisitions. Discover why big tech is buying developer ecosystems despite low enterprise adoption.
Anthropic has launched an initiative offering free API credits to researchers in biology and life sciences, supporting projects that aim to accelerate discovery through advanced AI tools like Claude Science. The program includes tiered funding and excludes certain regions due to export controls.
Stability AI has secured $76 million in a new Series B funding round, backed by major entertainment and gaming firms including Universal Music Group, Warner Music Group, Sony Music Group, and Electronic Arts. The investment reflects growing industry interest in co-developing generative AI tools using licensed content.
Google DeepMind releases Gemini 3.7 Flash, boosting coding and agent performance significantly while cutting introductory costs by half for developers.
Meta has released Muse Glimmer, a 30-billion parameter multimodal model optimized for local execution on consumer hardware using advanced quantization and speculative decoding techniques.
Discover Meta Muse Code, the new AI agent for large repositories. Built on Muse Spark 1.2, it offers parallel processing at a lower cost than competitors.
The Open Secure AI Alliance, formed by Nvidia and Microsoft, unites 37 tech leaders to develop open-source tools that protect against AI-driven cyber threats.
IBM's Granite multilingual embedding models (311 M & 97 M) achieve top retrieval scores under 100 M parameters, covering over 200+ languages.
Grabette is an open-source handheld system for affordable robot manipulation data collection, letting labs and hobbyists record and share datasets.
Thinking Machines Lab launches Inkling, the first open-weight multimodal model with 975 billion parameters and a “thinking effort” knob for speed-accuracy.
Discover NVIDIA Nemotron 3 Nano Omni, a 30B open multimodal model unifying vision, audio, and language for faster, efficient AI agent reasoning.
Avoid the scaling trap. Discover why open-source AI is the smarter, cost-effective choice for solo devs and startups compared to closed-source APIs.
Discover why the hybrid AI strategy wins in 2026. Compare open-source LLMs like Llama 4 and proprietary models like GPT-5 for cost and reasoning.
