Scoutari

AI builder 的決策情報台。

瀏覽

  • 今日情報
  • Skill 榜單
  • AI 週報
  • 關於與編輯方針
  • RSS

主題

  • 模型發布
  • API 與價格
  • MCP 與 Skills
  • AI Coding
  • 中國模型
  • 開源模型
  • Agent
  • Infra / 成本
  • 值得試用
© 2026 Scoutari · scoutari.com摘要由 AI 輔助生成,並始終附上原文連結。
情報收藏週報設定
SScoutariAI Builder Intel · decision desk
Skill RadarSavedPersonalizeSubmit sourceLog inSign up
Sign up
Skill RadarSavedPersonalizeSubmit sourceLog in

AI hardware

AI chips and accelerators: NVIDIA, TPUs, ASICs, and the semiconductor supply chain.

Model releasesAPI & pricingMCP & SkillsAI codingChinese modelsOpen sourceAgentsInfra & costMultimodal & imageAI videoVoice AIRobotics & embodiedAI safetyResearch & papersEnterprise AIAI fundingPolicy & regulationWorth trying
  • Sun, July 5Tools

    Anthropic Launches Claude Science Beta: Multi-Agent Workbench for Reproducible Pipelines

    On June 30, 2026, Anthropic released Claude Science Beta, a multi-agent collaboration workbench built on Claude models for reproducible pipelines in genomics, proteomics, and cheminformatics.

  • Sat, July 4Research

    NVIDIA AI Launches ASPIRE: Self-Improving Robot Framework with 31% Zero-Shot on LIBERO-Pro Long Tasks

    NVIDIA AI introduces ASPIRE, a framework that writes and improves robot control programs, distilling verified fixes into a reusable skill library.

  • Fri, July 3Research

    Anthropic Reportedly in Talks with Samsung for Custom AI Chips, But Says Nvidia Still Indispensable

    Anthropic is reportedly negotiating with Samsung to manufacture custom AI chips and has hired chip engineers, but stresses that Nvidia remains essential.

  • Thu, July 2Tools

    Anthropic in Talks with Samsung for Custom AI Chip Manufacturing

    Anthropic is negotiating with Samsung Electronics to be its manufacturing partner for custom AI chips.

  • Wed, July 1Model/API

    NVIDIA Releases Nemotron-Labs-TwoTower Diffusion Language Model

    NVIDIA released an open-weight diffusion language model built on a frozen autoregressive backbone.

  • Wed, July 1Model/API

    Nvidia rival Etched secures $1B in AI chip sales at $5B valuation

    Nvidia competitor Etched signs over $1B in contracts for its inference systems, reaching a $5B valuation.

  • Wed, July 1Model/API

    OpenAI Reportedly Slashes Inference Costs by Over Half for ChatGPT

    OpenAI reduced AI model inference costs by over half, cutting Nvidia GPU requirements to hundreds for ChatGPT.

  • Tue, June 30Tools

    NVIDIA Proposes 3 Workflows Using Synthetic Data & Fine-Tuning to Boost Visual AI Agent Accuracy

    NVIDIA blog introduces three workflows that leverage synthetic data and fine-tuning to improve visual AI agent accuracy.

  • Tue, June 30Research

    Three Workflows to Boost Visual AI Agent Accuracy with Synthetic Data and Fine-Tuning

    NVIDIA introduces three workflows using synthetic data and fine-tuning to enhance visual AI agent accuracy.

  • Tue, June 30Tools

    NVIDIA BioNeMo Agent Toolkit Transforms Biomolecular Models into AI Agent Skills

    NVIDIA open-sources BioNeMo Agent Toolkit to convert biomolecular models into callable skills for AI agents, boosting task completion from 57.1% to 100% in tests.

  • Sat, June 27Tools

    Building SFT Data from NVIDIA Open-SWE-Traces: Trace Parsing, Patch Analysis, Token Budget, Tool Usage Metrics

    A tutorial on streaming the NVIDIA Open-SWE-Traces dataset from Hugging Face in Google Colab to efficiently process agentic software engineering traces and generate a subset for fine-tuning.

  • Sat, June 27Research

    OpenAI, SpaceX and More Build Custom Chips to Challenge Nvidia's Dominance

    OpenAI, Google, Apple, and SpaceX are developing custom chips to reduce dependence on Nvidia, diversifying supplier risk.

  • Fri, June 26Model/API

    Optimize Model Training with NVIDIA Blackwell on Amazon SageMaker AI

    Configure training jobs on Amazon SageMaker AI to leverage Blackwell architecture advantages.

  • Wed, June 24Model/API

    OpenAI unveils first self-developed chip Jalapeño, made by Broadcom

    OpenAI launches its first self-designed chip Jalapeño, tailored for inference systems.

  • Wed, June 24Model/API

    OpenAI & Broadcom Launch Custom LLM Inference Chip 'Jalapeño'

    OpenAI and Broadcom introduced 'Jalapeño', a custom chip optimized for large language model inference, expected to run at scale by 2026.

  • Wed, June 24Research

    DFlash Speculative Decoding: Parallel Token Blocks Boost Throughput Up to 15x on NVIDIA Blackwell

    UC San Diego's DFlash replaces autoregressive draft generation with a lightweight block diffusion model, generating entire token blocks in a single forward pass for accelerated speculative decoding.

  • Wed, June 24Model/API

    OpenAI & Broadcom unveil Jalapeño chip for LLM inference

    OpenAI and Broadcom have introduced a custom AI chip called Jalapeño, optimized for large language model inference to boost performance, efficiency, and scale.

  • Wed, June 24Model/API

    Python Tutorial for ASR, Translation, and SRT Export Using NVIDIA Canary-1B-v2

    A tutorial on building multilingual ASR and speech translation pipelines in Python with NVIDIA Canary-1B-v2, including SRT subtitle export.

  • Tue, June 23Tools

    Nvidia Rubin Liquid-Cooled Data Center Design Promises to Slash Water and Power Use

    Nvidia claims its Rubin-generation fully liquid-cooled data center reference design drastically reduces electricity use and nearly eliminates water consumption.

  • Tue, June 23Tools

    SpaceX Signs Compute Deal with Open-Source AI Lab Reflection AI

    SpaceX will provide Nvidia GB300 AI chips and hardware access to Reflection AI for $150 million monthly.

  • Sat, June 20Agent

    NVIDIA AI launches SpatialClaw: a training-free agent using code as spatial reasoning action interface

    NVIDIA AI introduces SpatialClaw, a training-free agent that writes Python code in a persistent kernel to compose perception tools for 3D spatial reasoning.

  • Thu, June 18Tools

    NVIDIA SkillSpector Tutorial: Scan AI Skill Security Risks with Static Analysis and SARIF Reports

    This article introduces using NVIDIA SkillSpector to scan AI skills for security risks before deployment.

  • Tue, June 16Tools

    Microsoft Launches Surface Laptop 8 and Surface Pro 12 with Snapdragon X2 Chip

    Microsoft launches new Surface Laptop 8 and Surface Pro 12 powered by Qualcomm Snapdragon X2 processors.