What You Should Know Academic medical center UCSF Health, venture capital giant Kleiner Perkins, and Doerr Capital have announced the launch of UCSF Health Converge, a dedicated health AI accelerator built directly inside a top-ranked academic health network. Rejecting the isolated development models of traditional tech hubs, the accelerator centers on an “inside-out” approach, placing…
TerraFirma Inc., a startup with a platform that enables construction teams to remotely operate heavy machinery, has raised $115 million in funding. The bulk of the capital arrived in the form of a $100 million Series A round led by Kleiner Perkins. TerraFirma disclosed today that Bain Capital, Definition and more than a half-dozen other…
(This is a guest post by Developer Relations Engineer Thomas Steiner from the Chrome team at Google.) Transformers.js provides Web developers with a simple way to use the power of transformers in their Web apps through task-specific pipelines. To run inference in the browser, developers create an instance of pipeline() and specify a task they…
Reinforcement learning (RL) is central to aligning language models, from reinforcement learning with human feedback (RLHF) within AI assistants to newer reinforcement learning with verifiable rewards (RLVR) workflows for reasoning and agent tasks. RL is now becoming a practical technique for specialized AI where enterprises need more accurate agents for domain-specific workflows. Open models provide…
(A production RAG pipeline for PDFs: relational parsing, TOC retrieval, typed answers) we upgraded each of the four bricks: document parsing, question parsing, retrieval, and generation, and wired them into one clean, linear pipeline. A PDF goes in, and a typed, cited answer comes out. That was one paper and one question. The real test…
So much for the idea that open source cuts computing needs. Moonshot, the AI startup behind the Kimi K3 model, has temporarily stopped selling new subscriptions. Over the past 48 hours, “demand has pushed close to the limits of our current capacity,” the company announced on X about its recently released Kimi K3 model. Current subscribers aren’t affected, and…
This post is co-written with Vijay Venkatesh, CTO at Bluesight. If you build software for hospitals, you know that compliance work scales poorly. Hospitals managing 340B Drug Pricing Program compliance face a compounding data problem. Proving that a Group Purchasing Organization (GPO) purchased drug qualifies for an exception requires cross-referencing each purchase against several sources…
Thinking Machines Lab has unveiled Inkling, its first general-purpose open-weights foundation model. It is a multimodal MoE model with 975B parameters, 41B active parameters, and a 1M-token context window. Rather than chasing benchmark supremacy, Inkling is designed as a customizable foundation for multimodal reasoning, agentic AI, coding, tool use, audio and vision tasks, and domain-specific…
Getting the most out of your parameters Traditional scaling laws tell us that to achieve the best performance, we need to scale FLOPs, often with more parameters or data. But as models move to the edge and inference costs skyrocket, we wonder: Can we scale quality without inflating memory footprint? To that end, we’ve been…
Introduction Key–Value Cache (KV cache/KVCache) is a foundational building block of modern LLM serving systems. It stores past attention states so the model can generate new tokens efficiently without excessive re-computation. There are two phases to LLM inference: Prefill and Decode. In the Prefill phase, the attention states are computed for each token in the…