<?xml version="1.0" encoding="UTF-8"?><rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom"><channel><title>AI news, with the source in view</title><link>https://aisubhub.top/en/news/</link><description>Product releases, model updates and research explained with original sources and practical limits.</description><language>en</language><lastBuildDate>Tue, 20 Oct 2026 10:29:03 GMT</lastBuildDate><atom:link href="https://aisubhub.top/en/news/feed.xml" rel="self" type="application/rss+xml"/><item><title>Microsoft’s sovereign AI update: what is available, previewed or still ahead</title><link>https://aisubhub.top/en/news/microsoft-sovereign-ai-local-controls/</link><guid isPermaLink="true">https://aisubhub.top/en/news/microsoft-sovereign-ai-local-controls/</guid><pubDate>Tue, 20 Oct 2026 09:36:37 GMT</pubDate><description>The October 20 announcement connects local AI, development tools and operator controls, but mixes current features with previews and future Ignite releases.</description><source url="https://blogs.microsoft.com/blog/2026/10/20/advancing-digital-sovereignty-with-greater-control-and-choice/">Original source</source></item><item><title>BSC opens a selective route from Spanish AI startups to sovereign computing</title><link>https://aisubhub.top/en/news/bsc-ai-factory-inception-startup-access/</link><guid isPermaLink="true">https://aisubhub.top/en/news/bsc-ai-factory-inception-startup-access/</guid><pubDate>Tue, 20 Oct 2026 09:36:37 GMT</pubDate><description>The accelerator combines technical support and business mentoring; access, selection and early customer examples have important limits.</description><source url="https://blogs.nvidia.com/blog/spain-startups-barcelona-supercomputing-center/">Original source</source></item><item><title>AWS adds schema-constrained Claude outputs in Bedrock</title><link>https://aisubhub.top/en/news/bedrock-claude-structured-outputs/</link><guid isPermaLink="true">https://aisubhub.top/en/news/bedrock-claude-structured-outputs/</guid><pubDate>Tue, 20 Oct 2026 06:34:05 GMT</pubDate><description>Developer interfaces gain structured JSON and strict tool arguments, while factual checks and unsupported constraints remain application responsibilities.</description><source url="https://aws.amazon.com/blogs/machine-learning/structured-outputs-for-anthropic-claude-models-on-amazon-bedrock/">Original source</source></item><item><title>SageMaker Studio lowers the new-account barrier, with credit and GPU limits still attached</title><link>https://aisubhub.top/en/news/sagemaker-studio-new-account-onboarding/</link><guid isPermaLink="true">https://aisubhub.top/en/news/sagemaker-studio-new-account-onboarding/</guid><pubDate>Tue, 20 Oct 2026 10:29:03 GMT</pubDate><description>AWS connects Studio to a simpler sign-up flow; the useful change is faster entry, not unlimited cloud compute.</description><source url="https://aws.amazon.com/blogs/machine-learning/developers-can-now-access-sagemaker-studio-through-the-new-aws-experience/">Original source</source></item><item><title>Germany and NVIDIA outline industrial AI cooperation</title><link>https://aisubhub.top/en/news/nvidia-germany-industrial-ai-cooperation/</link><guid isPermaLink="true">https://aisubhub.top/en/news/nvidia-germany-industrial-ai-cooperation/</guid><pubDate>Tue, 20 Oct 2026 06:34:05 GMT</pubDate><description>NVIDIA describes an intent declaration and planned investments, with deployment outcomes still unproven.</description><source url="https://blogs.nvidia.com/blog/germany-wants-to-lead-the-world-in-industrial-ai-how-nvidia-will-help/">Original source</source></item><item><title>AWS details Pegasus 1.5 video analysis on Bedrock</title><link>https://aisubhub.top/en/news/pegasus-15-bedrock-video-metadata/</link><guid isPermaLink="true">https://aisubhub.top/en/news/pegasus-15-bedrock-video-metadata/</guid><pubDate>Tue, 20 Oct 2026 06:34:05 GMT</pubDate><description>The October 19 guide explains October 9 availability, timestamped JSON and video analysis up to two hours.</description><source url="https://aws.amazon.com/blogs/machine-learning/pegasus-1-5-from-twelvelabs-is-now-available-in-amazon-bedrock/">Original source</source></item><item><title>Deepgram&#39;s Trainium case separates model, kernel and serving costs</title><link>https://aisubhub.top/en/news/deepgram-nova3-trainium2-batch-inference/</link><guid isPermaLink="true">https://aisubhub.top/en/news/deepgram-nova3-trainium2-batch-inference/</guid><pubDate>Tue, 20 Oct 2026 09:36:37 GMT</pubDate><description>An AWS case study reports more offline Nova-3 audio per dollar after hardware-specific tuning, with limits on what the gains mean.</description><source url="https://aws.amazon.com/blogs/machine-learning/how-deepgram-optimized-voice-ai-at-scale-with-aws-trainium/">Original source</source></item><item><title>Two vLLM KV strategies target different latency bottlenecks</title><link>https://aisubhub.top/en/news/vllm-kv-transfer-latency-and-multi-turn-reuse/</link><guid isPermaLink="true">https://aisubhub.top/en/news/vllm-kv-transfer-latency-and-multi-turn-reuse/</guid><pubDate>Tue, 20 Oct 2026 09:36:37 GMT</pubDate><description>AWS&#39;s paired guides contrast push-based hand-off with bidirectional reuse, using separate benchmarks that should not be stacked.</description><source url="https://aws.amazon.com/blogs/machine-learning/push-based-kv-transfer-to-improve-ttft-and-tpot-for-disaggregated-inference-in-vllm/">Original source</source></item><item><title>NVIDIA links sovereign broadcasting plans with faster media search</title><link>https://aisubhub.top/en/news/ebu-sovereign-media-frameio-agentic-search/</link><guid isPermaLink="true">https://aisubhub.top/en/news/ebu-sovereign-media-frameio-agentic-search/</guid><pubDate>Tue, 20 Oct 2026 09:36:37 GMT</pubDate><description>The EBU initiative concerns governance and infrastructure; Adobe Frame.io performance claims describe a separate technical track.</description><source url="https://blogs.nvidia.com/blog/media-agentic-security-european-public-service-broadcasting-global-content-production/">Original source</source></item><item><title>Microsoft puts internal AI agents under continuous security scrutiny</title><link>https://aisubhub.top/en/news/microsoft-agentic-insider-2026-10-19/</link><guid isPermaLink="true">https://aisubhub.top/en/news/microsoft-agentic-insider-2026-10-19/</guid><pubDate>Tue, 20 Oct 2026 06:34:05 GMT</pubDate><description>Microsoft’s October 19 security argument focuses on the permissions and actions of agents already inside an organization.</description><source url="https://blogs.microsoft.com/blog/2026/10/19/securing-the-agentic-insider/">Original source</source></item><item><title>OpenAI reports progress on coding in less common languages</title><link>https://aisubhub.top/en/news/openai-low-resource-coding-evaluations/</link><guid isPermaLink="true">https://aisubhub.top/en/news/openai-low-resource-coding-evaluations/</guid><pubDate>Tue, 20 Oct 2026 06:34:05 GMT</pubDate><description>New evaluations examine SQLite implementations and behavior-preserving COBOL migration, with results reported by OpenAI.</description><source url="https://openai.com/index/low-resource-programming-languages/">Original source</source></item><item><title>Qwen Code v0.25.3 adds isolated review agents and clearer execution controls</title><link>https://aisubhub.top/en/news/qwen-code-v0253-2026-10-19/</link><guid isPermaLink="true">https://aisubhub.top/en/news/qwen-code-v0253-2026-10-19/</guid><pubDate>Tue, 20 Oct 2026 06:34:05 GMT</pubDate><description>The October 19 stable release focuses on managed sessions, tool configuration and workspace safeguards, rather than a new model.</description><source url="https://github.com/QwenLM/qwen-code/releases/tag/v0.25.3">Original source</source></item><item><title>Hugging Face outlines native Trainium support through TorchNeuron</title><link>https://aisubhub.top/en/news/hugging-face-torchneuron-2026-10-19/</link><guid isPermaLink="true">https://aisubhub.top/en/news/hugging-face-torchneuron-2026-10-19/</guid><pubDate>Tue, 20 Oct 2026 06:34:05 GMT</pubDate><description>The October 19 announcement describes direct integration with familiar libraries and an October 20 public-beta schedule.</description><source url="https://huggingface.co/blog/hf-torchneuron">Original source</source></item><item><title>RTX Spark launch joins Windows laptops with a phased local-AI software ecosystem</title><link>https://aisubhub.top/en/news/rtx-spark-windows-launch-software-stages/</link><guid isPermaLink="true">https://aisubhub.top/en/news/rtx-spark-windows-launch-software-stages/</guid><pubDate>Tue, 20 Oct 2026 10:29:03 GMT</pubDate><description>Three NVIDIA posts describe one product launch; hardware availability does not make every announced app feature final or free.</description><source url="https://blogs.nvidia.com/blog/local-ai-rtx-spark-windows-laptops-celebration/">Original source</source></item><item><title>DeepSeek refreshes Flash behind its existing API alias</title><link>https://aisubhub.top/en/news/deepseek-flash-1017-update/</link><guid isPermaLink="true">https://aisubhub.top/en/news/deepseek-flash-1017-update/</guid><pubDate>Tue, 20 Oct 2026 06:34:05 GMT</pubDate><description>The October 17 model update keeps API syntax and pricing unchanged; it is not a new subscription.</description><source url="https://api-docs.deepseek.com/updates/">Original source</source></item><item><title>AWS checks agent quality across nine lifecycle stages, not just an uptime dashboard</title><link>https://aisubhub.top/en/news/agentcore-nine-stage-quality-evaluation/</link><guid isPermaLink="true">https://aisubhub.top/en/news/agentcore-nine-stage-quality-evaluation/</guid><pubDate>Tue, 20 Oct 2026 10:29:03 GMT</pubDate><description>An internal sales-agent case connects domain-specific evaluation, release replay and live monitoring; its thresholds remain workload-specific.</description><source url="https://aws.amazon.com/blogs/machine-learning/building-trust-in-ai-agents-a-production-ready-evaluation-framework-with-amazon-bedrock-agentcore/">Original source</source></item><item><title>Anthropic studies changing work and junior hiring inside its own company</title><link>https://aisubhub.top/en/news/anthropic-internal-employment-study/</link><guid isPermaLink="true">https://aisubhub.top/en/news/anthropic-internal-employment-study/</guid><pubDate>Tue, 20 Oct 2026 06:34:05 GMT</pubDate><description>The internal report connects agent adoption with shifting work patterns, while explicitly limiting causal and economy-wide conclusions.</description><source url="https://www.anthropic.com/research/employment-at-anthropic">Original source</source></item><item><title>AWS&#39;s Reflection Beam guide separates active compute from deployment size</title><link>https://aisubhub.top/en/news/reflection-beam-aws-deployment-guide/</link><guid isPermaLink="true">https://aisubhub.top/en/news/reflection-beam-aws-deployment-guide/</guid><pubDate>Tue, 20 Oct 2026 09:36:37 GMT</pubDate><description>The October 16 walkthrough explains managed and HyperPod routes for a 501-billion-parameter coding model, without a universal price or performance guarantee.</description><source url="https://aws.amazon.com/blogs/machine-learning/introducing-reflection-beam-an-open-weight-model-for-agentic-coding-on-aws/">Original source</source></item><item><title>An a16z analysis examines why AI shopping still meets resistance</title><link>https://aisubhub.top/en/news/a16z-agentic-commerce-permissions/</link><guid isPermaLink="true">https://aisubhub.top/en/news/a16z-agentic-commerce-permissions/</guid><pubDate>Tue, 20 Oct 2026 06:34:05 GMT</pubDate><description>Moses Sternstein separates product research from delegated checkout, with explicit limits on the underlying evidence.</description><source url="https://www.a16z.news/p/this-week-in-charts-agents-denied">Original source</source></item><item><title>Cursor adds shareable cloud-agent canvases and a beta Environments API</title><link>https://aisubhub.top/en/news/cursor-cloud-agent-canvases-environments-2026-10-16/</link><guid isPermaLink="true">https://aisubhub.top/en/news/cursor-cloud-agent-canvases-environments-2026-10-16/</guid><pubDate>Tue, 20 Oct 2026 09:36:37 GMT</pubDate><description>The October 16 update combines interactive outputs with programmatic cloud-environment management and a larger repository limit.</description><source url="https://cursor.com/changelog/cloud-agent-canvases-and-environments">Original source</source></item><item><title>Elorian introduces Glimpse as a visual reasoning research preview</title><link>https://aisubhub.top/en/news/elorian-glimpse-research-preview/</link><guid isPermaLink="true">https://aisubhub.top/en/news/elorian-glimpse-research-preview/</guid><pubDate>Tue, 20 Oct 2026 06:34:05 GMT</pubDate><description>The launch describes API and chat access, while distinguishing recorded model outputs from illustrative reference answers.</description><source url="https://elorian.ai/blog/introducing-glimpse">Original source</source></item><item><title>Rosalind Academic Plan offers conditional access for US university labs</title><link>https://aisubhub.top/en/news/rosalind-academic-plan-eligibility/</link><guid isPermaLink="true">https://aisubhub.top/en/news/rosalind-academic-plan-eligibility/</guid><pubDate>Tue, 20 Oct 2026 06:34:05 GMT</pubDate><description>Eligible labs can apply for sponsored seats, but GPT-Rosalind requires approval and API access is excluded.</description><source url="https://openai.com/index/introducing-the-rosalind-academic-plan/">Original source</source></item><item><title>Microsoft Research presents Magnetar for proving ordinary program execution</title><link>https://aisubhub.top/en/news/microsoft-research-magnetar-2026-10-15/</link><guid isPermaLink="true">https://aisubhub.top/en/news/microsoft-research-magnetar-2026-10-15/</guid><pubDate>Tue, 20 Oct 2026 06:34:05 GMT</pubDate><description>The October 15 research report describes a RISC-V zero-knowledge virtual machine and carefully scoped benchmark gains.</description><source url="https://www.microsoft.com/en-us/research/blog/proof-based-general-purpose-computing-with-magnetar/">Original source</source></item><item><title>AWS’s Trust in Depth separates an agent’s decisions from the controls it cannot override</title><link>https://aisubhub.top/en/news/aws-trust-in-depth-agent-boundaries/</link><guid isPermaLink="true">https://aisubhub.top/en/news/aws-trust-in-depth-agent-boundaries/</guid><pubDate>Tue, 20 Oct 2026 10:29:03 GMT</pubDate><description>The four-layer approach links scope, identity, external enforcement and evidence, while limiting formal proof to explicit rules and assumptions.</description><source url="https://aws.amazon.com/blogs/machine-learning/trust-in-depth-giving-agents-more-autonomy-without-losing-control/">Original source</source></item><item><title>Nemotron 3.5 separates multimodal agents from small local deployments</title><link>https://aisubhub.top/en/news/nemotron-35-super-vl-bolt-release/</link><guid isPermaLink="true">https://aisubhub.top/en/news/nemotron-35-super-vl-bolt-release/</guid><pubDate>Tue, 20 Oct 2026 09:36:37 GMT</pubDate><description>Super VL, a 120B-parameter model, arrives alongside a 24GB Bolt variant; performance and availability claims depend on model and configuration.</description><source url="https://blogs.nvidia.com/blog/nemotron-3-5-super-vl/">Original source</source></item><item><title>NVIDIA&#39;s European science pledge combines open materials with conditional compute routes</title><link>https://aisubhub.top/en/news/nvidia-europe-science-access-pledge/</link><guid isPermaLink="true">https://aisubhub.top/en/news/nvidia-europe-science-access-pledge/</guid><pubDate>Tue, 20 Oct 2026 10:29:03 GMT</pubDate><description>The October 15 commitment offers learning resources and research pathways, not an automatic GPU allocation or a blanket licence.</description><source url="https://blogs.nvidia.com/blog/ec-scientific-computing-pledge/">Original source</source></item><item><title>NVIDIA&#39;s Nordic update separates installed systems from future AI capacity</title><link>https://aisubhub.top/en/news/nvidia-nordic-infrastructure-project-stages/</link><guid isPermaLink="true">https://aisubhub.top/en/news/nvidia-nordic-infrastructure-project-stages/</guid><pubDate>Tue, 20 Oct 2026 10:29:03 GMT</pubDate><description>The regional account links a shared supercomputer, application work and new support plans, without making them one universally available service.</description><source url="https://blogs.nvidia.com/blog/nordics-technology-ecosystem/">Original source</source></item><item><title>MAI-Base-2: how Microsoft separates training efficiency from model availability</title><link>https://aisubhub.top/en/news/mai-base-2-pretraining-methods/</link><guid isPermaLink="true">https://aisubhub.top/en/news/mai-base-2-pretraining-methods/</guid><pubDate>Tue, 20 Oct 2026 09:36:37 GMT</pubDate><description>Microsoft describes a 2.8T-parameter base model, repeated-data scaling and infrastructure changes. The reported efficiency does not establish an available chat product.</description><source url="https://microsoft.ai/news/frontier-pre-training-at-microsoft-ai/">Original source</source></item><item><title>AgentCore document processing reuses validated instructions, not newly trained model weights</title><link>https://aisubhub.top/en/news/agentcore-document-processing-instruction-reuse/</link><guid isPermaLink="true">https://aisubhub.top/en/news/agentcore-document-processing-instruction-reuse/</guid><pubDate>Tue, 20 Oct 2026 10:29:03 GMT</pubDate><description>AWS’s sample combines eight agents with business-rule validation; its faster second run and monthly estimate describe a narrow example.</description><source url="https://aws.amazon.com/blogs/machine-learning/build-self-improving-document-processing-using-amazon-bedrock-agentcore/">Original source</source></item><item><title>AgentCore’s presentation sample is a RAG workflow, not a four-cent slide service</title><link>https://aisubhub.top/en/news/agentcore-rag-presentation-generation/</link><guid isPermaLink="true">https://aisubhub.top/en/news/agentcore-rag-presentation-generation/</guid><pubDate>Tue, 20 Oct 2026 10:29:03 GMT</pubDate><description>AWS reports a six-slide internal test, with separate model cost, asynchronous delivery and an important OpenSearch network condition.</description><source url="https://aws.amazon.com/blogs/machine-learning/generate-presentations-with-amazon-bedrock-agentcore/">Original source</source></item><item><title>Tamnoon describes how Tami separates cloud investigation from approved repairs</title><link>https://aisubhub.top/en/news/tamnoon-bedrock-governed-cloud-remediation/</link><guid isPermaLink="true">https://aisubhub.top/en/news/tamnoon-bedrock-governed-cloud-remediation/</guid><pubDate>Tue, 20 Oct 2026 10:29:03 GMT</pubDate><description>An October 14 AWS case study explains four specialized agents, conservative safety scoring and controlled execution; reported improvements are customer-case claims.</description><source url="https://aws.amazon.com/blogs/machine-learning/how-tamnoon-built-ai-powered-cloud-security-remediation-with-amazon-bedrock/">Original source</source></item><item><title>AWS shows a repeatable JumpStart deployment and evaluation loop</title><link>https://aisubhub.top/en/news/sagemaker-jumpstart-repeatable-model-evaluation/</link><guid isPermaLink="true">https://aisubhub.top/en/news/sagemaker-jumpstart-repeatable-model-evaluation/</guid><pubDate>Tue, 20 Oct 2026 10:29:03 GMT</pubDate><description>The October 14 guide connects model discovery, sample requests and endpoint cleanup, while making clear that its timing example is not a general benchmark.</description><source url="https://aws.amazon.com/blogs/machine-learning/deploy-and-evaluate-models-with-amazon-sagemaker-jumpstart/">Original source</source></item><item><title>NVIDIA plans investments in five UK startups, spanning biology to quantum systems</title><link>https://aisubhub.top/en/news/nvidia-uk-startup-planned-investments-2026/</link><guid isPermaLink="true">https://aisubhub.top/en/news/nvidia-uk-startup-planned-investments-2026/</guid><pubDate>Tue, 20 Oct 2026 10:29:03 GMT</pubDate><description>The October 14 announcement names new investment intentions alongside existing deployments, without disclosing transaction amounts or completed funding.</description><source url="https://blogs.nvidia.com/blog/uk-ecosystem-deployment/">Original source</source></item><item><title>European city AI projects show a mix of deployments, pilots and cost targets</title><link>https://aisubhub.top/en/news/nvidia-european-city-agents-project-stages/</link><guid isPermaLink="true">https://aisubhub.top/en/news/nvidia-european-city-agents-project-stages/</guid><pubDate>Tue, 20 Oct 2026 10:29:03 GMT</pubDate><description>NVIDIA describes local agents and digital twins, but the status and evidence differ sharply between municipalities and ports.</description><source url="https://blogs.nvidia.com/blog/european-cities-agents-urban-operations/">Original source</source></item><item><title>Bedrock private offers need active grants before negotiated pricing applies</title><link>https://aisubhub.top/en/news/bedrock-private-offer-managed-entitlements/</link><guid isPermaLink="true">https://aisubhub.top/en/news/bedrock-private-offer-managed-entitlements/</guid><pubDate>Tue, 20 Oct 2026 10:29:03 GMT</pubDate><description>AWS contrasts a reviewed one-time script with seller-allowlisted automation for multi-account private offers; neither creates a public discount or instant billing confirmation.</description><source url="https://aws.amazon.com/blogs/machine-learning/manage-private-pricing-offers-for-amazon-bedrock-with-managed-entitlements/">Original source</source></item><item><title>Orca explains how a smaller schema context improved Discovery AI queries</title><link>https://aisubhub.top/en/news/orca-discovery-ai-schema-aware-security-queries/</link><guid isPermaLink="true">https://aisubhub.top/en/news/orca-discovery-ai-schema-aware-security-queries/</guid><pubDate>Tue, 20 Oct 2026 10:29:03 GMT</pubDate><description>The October 13 case study combines deterministic matching, retrieved examples and validation; its 900-plus “models” are cloud resource schemas, not language models.</description><source url="https://aws.amazon.com/blogs/machine-learning/natural-language-security-queries-at-orca-security-using-amazon-bedrock/">Original source</source></item><item><title>Gemini 4 Argon arrives through distinct subscription and API channels</title><link>https://aisubhub.top/en/news/gemini-4-argon-access-channels/</link><guid isPermaLink="true">https://aisubhub.top/en/news/gemini-4-argon-access-channels/</guid><pubDate>Tue, 20 Oct 2026 06:34:05 GMT</pubDate><description>Google’s October 13 announcement separates Pro and Ultra app access from paid API usage and introductory token prices.</description><source url="https://blog.google/innovation-and-ai/models-and-research/gemini-models/introducing-gemini-4-argon/">Original source</source></item><item><title>ShipStation’s Deckhand case study separates answering IT questions from taking action</title><link>https://aisubhub.top/en/news/shipstation-deckhand-agentcore-internal-helpdesk/</link><guid isPermaLink="true">https://aisubhub.top/en/news/shipstation-deckhand-agentcore-internal-helpdesk/</guid><pubDate>Tue, 20 Oct 2026 10:29:03 GMT</pubDate><description>The October 13 account describes a Slack help desk agent with confirmed ticket creation, curated knowledge and feedback; its 97.5% figure measures replies, not universal resolution.</description><source url="https://aws.amazon.com/blogs/machine-learning/how-shipstation-global-built-an-ai-it-help-desk-on-amazon-bedrock-agentcore/">Original source</source></item><item><title>Why Anthropic built Scion for an agent-scale monorepo</title><link>https://aisubhub.top/en/news/anthropic-scion-agent-scale-source-control-2026-10-13/</link><guid isPermaLink="true">https://aisubhub.top/en/news/anthropic-scion-agent-scale-source-control-2026-10-13/</guid><pubDate>Tue, 20 Oct 2026 09:36:37 GMT</pubDate><description>The internal source-control system moves computation to servers and retrieves immutable objects on demand; the engineering account is a workload-specific case study, not a general replacement for Git.</description><source url="https://www.anthropic.com/engineering/redesigning-source-control">Original source</source></item><item><title>Claude Haiku 5.5 targets high-volume work with tiered API pricing</title><link>https://aisubhub.top/en/news/claude-haiku-5-5-api-pricing/</link><guid isPermaLink="true">https://aisubhub.top/en/news/claude-haiku-5-5-api-pricing/</guid><pubDate>Tue, 20 Oct 2026 06:34:05 GMT</pubDate><description>Anthropic’s October 7 announcement sets different token prices above and below a 100,000-token prompt threshold.</description><source url="https://www.anthropic.com/claude-haiku-5-5">Original source</source></item></channel></rss>