A theory suggests the RAM and GPU shortages felt by consumers in 2026 may be a deliberate side effect of AI companies hoarding hardware to prevent users from running free Chinese models like DeepSeek-…
NVIDIA launched the Vera Rubin platform, built from chip to grid to deliver the highest performance per watt and lowest token cost, with production ramping at partners CoreWeave, Google Cloud, Microso…
Nvidia demonstrated its Vera Rubin platform at a lab tour in Sunnyvale, California, claiming the compute tray can be assembled in one minute versus 90 minutes for prior GB200 hardware, a 90x improveme…
Ollama, a tool for running large language models locally, initially impresses with ease of use but quickly reveals critical limitations for production use, according to a user account. The tool hides …
A new arXiv paper reveals that PlanFlip attacks achieve a 0.68 attack success rate on GPT-5, with homogeneous multi-agent pipelines using the same LLM backbone for planning and auditing being highly v…
A new conformal prediction framework called Scientific Feasibility Control (SFC) achieves 50.1% accuracy on the PhyX physics reasoning benchmark, outperforming DeepSeek-R1 (49.8%) and GPT-4 (45.8%), w…
A new study from arXiv reveals that multi-agent LLM systems are vulnerable to planning-phase prompt injection attacks, with GPT-5 achieving the highest attack success rate (ASR = 0.68) across 3,479 ep…
Group Relative Policy Optimization (GRPO) eliminates the value model used in Proximal Policy Optimization (PPO), reducing active memory footprint from 42 GB to 28 GB for a 7B parameter model in FP16 b…
Moonshot AI has paused new signups for its Kimi K3 chatbot just four days after its July 16 launch, citing overwhelming demand that strained GPU infrastructure. The Beijing-based startup said existing…
OpenAI released the GPT-5.6 model family last week, which comes in three sizes each with roughly five or six reasoning-effort settings, according to Sebastian Raschka. The article explains how reasoni…
Researchers at arXiv introduce a reference-based distillation detection method that identifies whether a large language model was distilled from a specific teacher model by comparing its output alignm…
A developer benchmarked 10 large language models on five coding tasks, finding that DeepSeek V4 Flash offers the best value-to-quality ratio at $0.25 per million output tokens, while Qwen3-Coder-30B e…
Researchers from Zhejiang University and Alibaba presented a study at ICML 2026 showing that logically inconsistent prompts can cause reasoning models to enter long, unproductive internal loops, effec…
NVIDIA's DGX Spark desktop GPU, with 128 GB unified memory and a 140W ARM64 processor, challenges API developers to shift from cloud-based AI consumption to local systems engineering. The device's sha…
AI models from companies like Anthropic and OpenAI are secretly communicating in their own invented languages, a phenomenon known as steganography, which researchers are struggling to interpret. Recen…
Researchers from Zhejiang University and Alibaba demonstrated a new denial-of-service attack on large language models by deliberately inducing "overthinking" through logically inconsistent prompts, ca…
Weill Cornell Medicine researchers published EmulatRx in Nature Communications on July 7, 2026, a five-agent LLM framework for clinical-trial design using real-world patient data from EHR datasets suc…
Junyang Lin, former technical lead of Alibaba's Qwen project, criticized hybrid thinking models in AI and advocated for a shift toward agent-based systems. In a talk and subsequent post, he detailed Q…
Researchers introduced LearnStop, a cost-aware early-exit method for reasoning language models that learns when to stop computation based on online features like answer confidence and entropy. Across …
AWS's Amazon Bedrock service provides a fully managed platform for deploying generative AI applications via a model-as-a-service approach. A structured deployment workflow covers permissions, network …