GLM-5.3 오픈 웨이트, Claude 자율 얼라인먼트, Gemini 과학 실험, 에이전트 실행 인프라

GLM-5.3 is now open-weight. Our most capable model for agentic coding and cyber defense is now available to download, run, and customize. Weights: https:// huggingface.co/zai-org/GLM-5.3 Tech blog: https:// z.ai/blog/glm-5.3

GLM-5.3 can now be run locally! The 2-bit model retains ~81% accuracy after we shrunk it from 1.51TB to 239GB (-83% size). Run on a 256GB Mac or RAM/VRAM setups. GLM-5.3 is the strongest open model to date. Guide: https:// unsloth.ai/docs/models/gl m-5.3 … GGUF: https:// huggingface.co/unsloth/GLM-5. 3-GGUF …

New Fellows Research: Can Claude autonomously align other AIs? We gave Claude 48 hours and 1 GPU to improve the alignment of small models. It researched and proposed methods, then trained and tested the models on its own. It worked surprisingly well.

Could a model one day align its stronger successors? As a first test, we had Sonnet 5 post-train an early checkpoint of Opus 4.8, a more capable model. It reached safety scores approaching those of production Opus 4.8, which went through our full alignment training.

Today, we're excited to share early progress on using Gemini to accelerate scientific discovery in the real-world. We present an extension of Co-Scientist which we use to collaborate with scientists across materials science, biology, and computer science.

New paper in Nature Communications! We show you can map out urban flooding by using vision-language models to detect floods in large-scale street scene datasets. In New York City, our method identifies flooded neighborhoods, home to 100k people, that current methods miss. 1/

Here’s what launched this week: — Gemini 3.5 Transcribe, our most precise speech-to-text model yet, designed to deliver intelligent transcriptions — Gemini Omni 1.1 Flash, bringing expanded creative capabilities and controls for video generation and editing — @GeminiApp ’s Live experience, moving beyond conversation to complex tasks with new features like Daily Brief, Gemini Spark, Personal Intelligence, and @Gmail inbox management — Expert Intelligence, a new cross- @Google initiative that lets you engage with and combine insights from trusted sources, starting with eligible @GooglePlay ebooks in @Gemini_Notebook

We’re rolling out Gemini Omni 1.1 Flash to make generative video highly controllable, faster to iterate on, and more polished for production-grade use. Here’s how you can try it in @FlowbyGoogle and more → https:// goo.gle/4zHEzJ2

LangChain OSS

@LangChain_OSS

1일 전

new MCP spec support is coming soon in @LangChain (and deepagents) try it out and let us know what you think!

The DeepSeek harness is engineering at its finest. 202K stars on GitHub. Damn! Clearly, they built this harness with future AI needs and capabilities in mind. Everything is a plugin and customizable, which is exactly how harnesses of the future need to be.

You can now resume your terminal sessions in the Claude Code desktop app. Type /resume to pick any session you started from the CLI. The session continues in the app with the full conversation and context intact.

Aravind Srinivas

@AravSrinivas

1일 전

Decagon is powering customer support for some of the biggest brands in the industry: Delta Airlines, Ticketmaster, Deutsche Telekom, American Airlines. A lot of support related questions require accurate online information and we’re happy to power that with Perplexity’s search.

Static embeddings offer unmatched throughput, but they also suffer from accuracy loss compared to traditional embedding models. Can we make them better for retrieval? We tried several things: raw maxsim scoring on the per-token embeddings training a small adapter model changing the distillation training target and teacher While none of these saw the results we wanted, it's an excellent dive into static embeddings and how they do (and don't) work! Blog: https:// llamaindex.ai/blog/exploring -static-embedding-retrieval …

Logan Markewich

@LoganMarkewich

1일 전

Obligatory plug for static embeddings. Are you using static embeddings in your projects? Would love to hear how people are leveraging them, whether for retrieval and beyond.

Google Gemma

@googlegemma

1일 전

Gemma 4 E2B running fully locally on an 8GB NVIDIA Jetson Orin Nano. Using native image/audio input and function calling, this edge setup is capable of vision, voice interaction, and camera control, all within an 8GB memory footprint!

Things getting wild in automated systems engineering. This new work claims to build an AI system that designed and deployed a frontier accelerator. Two human architects wrote the hardware specification, and an AI system generated everything below it. Performance model, RTL, UVM verification environments, formal proofs, firmware, and kernels, all in under TWO weeks with no human intervention below the spec. Every block reached 95 percent coverage through commercial EDA tools, a proprietary formal engine, and hardware-in-the-loop validation. The resulting accelerator, Redwood, targets single-batch low-latency inference. Projected onto Samsung 8nm it delivers 1.75x the throughput at 1.9x lower power against a measured Jetson Orin Nano, a 3.4x gain in performance per watt. The FPGA variant runs multi-billion-parameter Llama and Qwen models. They claim that it takes 48 hours for a specification change to reverify and redeploy to hardware. Paper: https:// arxiv.org/abs/2608.26418 Chat with Paper: https:// academy.dair.ai/papers/an-ai-s ystem-designed-and-deployed-a-frontier-accelerator-2608.26418 …