Auto Added by WPeMatico
Long-context inference makes the KV cache one of the main costs of serving LLMs. During autoregressive decoding, the cache grows with context length, batch size, and model depth. At high batch sizes and long contexts with 100K tokens across dozens of concurrent requests the KV cache consumes a large fraction of GPU memory. Compressing it […]
For years, authentication on the web followed one design assumption: a human sits behind a browser. Click a button. Fill out a form. Verify an email. Copy an API key and paste it somewhere else. That model does not work when the user is delegating work to an agent. Agents are already writing code, opening
WorkOS Releases auth.md: An Open Agent Registration Protocol Built on OAuth Standards Read More »
StepFun, the Shanghai-based AI lab, released StepAudio 2.5 Realtime. It is an end-to-end real-time speech large language model with fully customizable persona capabilities. StepAudio 2.5 Realtime is a voice model that operates in real time. Unlike pipeline-based systems that separate speech recognition, reasoning, and synthesis into sequential steps, this is an end-to-end model. Audio goes
Most web agents today drive a browser one action at a time. The model receives the current page state — as a screenshot or DOM text — and predicts the next click, keypress, or scroll. This action-at-a-time design made sense when language models had limited reasoning ability. As models have become more capable at writing
Linear attention replaces the unbounded KV cache of softmax attention with a fixed-size recurrent state. This cuts sequence mixing to linear time and decoding to constant memory. The hard part is not what to forget. It is how to edit a compressed memory without scrambling existing associations. NVIDIA has released Gated DeltaNet-2, a linear attention
Tencent has released TencentDB Agent Memory, an open-source memory system for AI agents. The project ships under the MIT license. It targets a problem familiar to anyone shipping long-horizon agents: context bloat and recall failure. It is symbolic short-term memory along with layered long-term memory. It integrates with OpenClaw as a plugin and with the
Instruction-tuned language models refuse harmful requests. But which part of the model is actually responsible — and how does that mechanism get installed during training? A new research from Nous Research team takes a neuron-level look at this question. The Nous research team developed contrastive neuron attribution (CNA), a method that identifies the specific MLP
Attackers increasingly target the packages, editor extensions, and AI tool configs on developer machines and not just production systems. Perplexity has open-sourced an internal tool it uses to address this problem. Perplexity released Bumblebee on GitHub. The tool is a read-only inventory collector for macOS and Linux developer endpoints. It is written entirely in Go
Microsoft Research’s AI Frontiers lab released Fara1.5. It is a family of computer-use agent (CUA) models for the browser. The release ships three sizes: Fara1.5-4B, Fara1.5-9B, and Fara1.5-27B. The models are integrated with MagenticLite, Microsoft’s sandboxed browser interface for these agents. Computer-use agents are pixel-to-action models that drive a real browser. They read screenshots and
Most AI models today are not designed for sustained, multi-step autonomous execution. Tasks like running hundreds of iterative code modifications, or chaining tool calls across hours without human intervention, require a different kind of model architecture and training focus. Alibaba’s Qwen team formally announced Qwen3.7-Max at the 2026 Alibaba Cloud Summit on May 20. Although,
Qwen Introduces Qwen3.7-Max: A Reasoning Agent Model With a 1M-Token Context Window Read More »
