RT afra wangthing i know about http://z.ai, the company behind GLM:1. http://z.ai is known as "Zhipu" before being rebranded as a sleeker "http://z.ai". The Chinese name…
.@cartesia runs one of the hardest inference workloads: real-time voice.Their stack has to keep long-lived streams moving, serve millions of audio minutes a day, and hold…
OpenAI is using AI to help the open source community better protect itself.
Re Sakana Fugu Technical Reporthttps://github.com/SakanaAI/fugu/blob/main/Fugu_technical_report.pdfRelease Notes: https://sakana.ai/fugu-release/
In a world of slop, creatives need an aesthetics-focused model. Here's how to work with it.
When annotators disagree on a label, the disagreement itself carries signal—and the number of annotators needed to capture it depends on the evaluation metric. We fine-tune…
LLM-as-a-judge panels aggregate votes from multiple models, with the expectation that diverse models yield more reliable evaluations. We develop a framework to measure the true informational…
ParallelKernelBench tests whether LLMs can write fast multi-GPU CUDA kernels across 87 real workloads. The best model solves under a third, but a few generated kernels…
Discover how Omio uses OpenAI to power conversational travel experiences, accelerate product development, and transform into an AI-native company.
Prompt Injection as Role Confusion First, I absolutely love this: This is a blog-style writeup of the paper. I wish every paper would come with one…
Sakana launched Fugu multi-agent orchestration; OpenAI shipped GPT-5.5-Cyber + Patch the Planet; Anthropic + Micron co-design memory; SpaceX inked $150M/month with Reflection AI.
This morning on Hacker News I saw Moebius: 0.2B Lightweight Image Inpainting Framework with 10B-Level Performance, describing a small but effective inpainting model - a model…
I work in firmware adjacent to AI, so not an ML guy exactly, so that's why I've come here. For work we got a bit concerned…
You can now connect Grok to Interactive Brokers, delivering high quality, up-to-date information on your portfolioInteractive Brokers: Interactive Brokers integrates with Grok. Research investments and analyze…
Public pushback against data centers has emphasized their water and energy consumption, and now Nvidia is highlighting its claim that the Rubin generation reference design for…
Article URL: https://thomasliao.com/eval-startups Comments URL: https://news.ycombinator.com/item?id=48637868 Points: 37 # Comments: 33
RT Pamela FoxThe @aiDotEngineer schedule wins for having the most possible ways that developers can interact with it: JSON, llms.txt, MCP server, iCal, embeddings, agent skillhttps://ai.engineer/worldsfair/scheduleNow…
RT Luma | Dream LabImmerse into the world of Blake Johnson’s imagination 🤯🚀Blake used Luma agent to compose the environment, characters, and VFX for a short…
We would often like to get a qualitative sense of a target model’s behaviors in important distributions (e.g. deployment, RL training, or evals). For example, we…
Patch the Planet: using frontier AI and working with professional security researchers to secure critical OSS projectsOpenAI: Patch the Planet is our effort to help open…
RT E2BOur own Matt Brockman is hosting a workshop on top 5 sandbox failure modes and how to address them at AI Engineer!Catch his workshop to…
US autoworkers union warns of robot automation as dark factory future looms.
RT Wing VCNew listen: @GraySwanAI cofounders @zicokolter and Matt Fredrikson sat down with @swyx on @latentspacepod to unpack why AI security is its own discipline; not…
Hi all, I'm looking for literature on relatively specific tooling. In autoregressive LLMs, there is substantial published work that used NLI on sub-claims produced by LLMs…