https://modelsatwork.news/article/nous-research-arete-27b-qwen-fine-tune-cuts-propaganda-rate-22-to-0-9-percent Models at Worken 2026-11-07T16:58:00.000Z Nous Research's Arete-27B cuts Qwen's propaganda rate from 22% to 0.9% Open-weight models, Fine-tuning, Evals, Model selection, Model release https://modelsatwork.news/article/vals-ai-cyberoffense-open-weight-deepseek-72-8-percent-vs-restricted-daybreak-80-8 Models at Worken 2026-11-07T16:20:00.000Z Vals AI: open DeepSeek model nears OpenAI's restricted cyber model Security, Benchmarks, Open-weight models, Agent cost, Evals https://modelsatwork.news/article/hugging-face-safeagents-proxy-blocks-agent-reward-hacking-tool-calls Models at Worken 2026-11-07T14:58:00.000Z Hugging Face's safeagents proxy blocks agent cheating before tools run Agent safety, Coding agents, Open source, Fine-tuning, Agent evaluation https://modelsatwork.news/article/nist-nsa-cisa-ten-practices-agent-control-incidents Models at Worken 2026-11-07T13:58:00.000Z NIST, NSA and CISA publish ten practices for agent control incidents Security, Agent safety, Standards, Monitoring, Egress controls https://modelsatwork.news/article/into-the-unknown-nine-ai-agents-fail-national-lab-research-bar Models at Worken 2026-11-07T11:55:00.000Z Nine AI agents tried lab follow-up research; none met the experts' bar Research, Benchmarks, Agent evaluation, Verification, Agents https://modelsatwork.news/article/amazon-quick-four-questions-trigger-agent-skill-workflow Models at Worken 2026-11-07T10:57:00.000Z Amazon Quick: new agent job? Often it's a trigger, not a new agent Agents, AWS, Vendor-published, Agent governance, Decision models https://modelsatwork.news/article/anthropic-agent-memory-let-model-judge-what-to-save Models at Worken 2026-11-07T09:58:00.000Z Anthropic: let the model pick what to remember, not fixed rules Context management, Agents, Evals, Anthropic, Managed Agents https://modelsatwork.news/article/dharma-field-extractor-2b-document-to-json-beats-opus-5-vendor-benchmark Models at Worken 2026-11-07T01:55:00.000Z Dharma's 2B Field Extractor turns documents into JSON for $2.24 an hour Small models, Document processing, Benchmarks, Self-hosting, Cost, Vendor-published https://modelsatwork.news/article/sagemaker-gpu-partitioning-mig-h100-slice-vs-a100 Models at Worken 2026-11-07T00:55:00.000Z SageMaker adds GPU partitioning: an H100 slice beat a whole A100 Amazon SageMaker AI, GPU, Inference, Cost, Multi-tenancy, Vendor-published https://modelsatwork.news/article/amazon-sort-center-mcp-server-data-lake-agents-fargate Models at Worken 2026-11-06T23:55:00.000Z Amazon sort centers put one MCP server between AI agents and the data lake MCP, Agents, Access control, Amazon Bedrock, AWS, Vendor-published https://modelsatwork.news/article/aws-okta-cross-app-access-agentcore-which-agent-acts-for-which-user-measured Models at Worken 2026-11-06T23:00:00.000Z AWS and Okta test which AI agent may act for which user Agent governance, Access control, Amazon Bedrock AgentCore, Agent security, Vendor-published https://modelsatwork.news/article/aws-reference-design-claude-code-remote-sessions-bedrock-agentcore Models at Worken 2026-11-06T22:05:00.000Z AWS sample runs Claude Code in isolated sessions, returns pull requests Coding agents, Amazon Bedrock AgentCore, Claude, Agent security, Vendor-published https://modelsatwork.news/article/taxbit-assist-cited-permission-aware-agent-under-20-cents-per-conversation Models at Worken 2026-11-06T21:05:00.000Z Taxbit Assist: a cited, permission-aware agent at under 20 cents a chat Case study, Amazon Bedrock AgentCore, RAG, Access control, Vendor-published https://modelsatwork.news/article/ricursive-ripple-65b-preview-open-weight-dense-model-terminal-bench-apache Models at Worken 2026-11-06T20:05:00.000Z Ricursive's Ripple-65B: Apache 2.0 open model that fits one DGX Spark Model release, Open-weight models, Coding agents, Self-hosting, Benchmarks https://modelsatwork.news/article/perplexity-teammates-always-on-agents-own-ongoing-work-max-200-dollars Models at Worken 2026-11-06T19:05:00.000Z Perplexity Teammates: always-on agents for ongoing work, on Max Agents, Personal agents, Access control, Pricing, Vendor-published https://modelsatwork.news/article/zalando-agentic-identity-broker-scoped-revocable-agent-access-show-hn Models at Worken 2026-11-06T18:10:00.000Z Zalando open-sources a broker that gives AI agents scoped, revocable access Agents, MCP, Access control, Open source, Show HN https://modelsatwork.news/article/greenlight-financial-luna-soc-agents-90-percent-cases-no-human-aws-vendor-published Models at Worken 2026-11-06T17:05:00.000Z Greenlight's SOC agents resolve 90% of 4,000 monthly cases unaided Multi-agent, Security, Human in the loop, Amazon Bedrock AgentCore, Vendor case study https://modelsatwork.news/article/liquid-ai-lfm3-2b-open-weight-models-decision-models-vendor-published Models at Worken 2026-11-06T15:05:00.000Z Liquid AI LFM3: six 2B open-weight models, from text to decisions Model release, Open-weight models, Decision models, Self-hosting, Vendor-published https://modelsatwork.news/article/rivet-open-source-agent-actors-s3-72-kb-vs-512-mb-sandbox-launch-hn Models at Worken 2026-11-06T13:05:00.000Z Rivet: open-source actors for agents, 72 KB each versus a 512 MB sandbox Agents, Open source, Self-hosting, Developer tools, Cost https://modelsatwork.news/article/claude-code-user-agent-email-address-issue-107757 Models at Worken 2026-11-06T12:05:00.000Z Claude Code users report their email in User-Agent headers Claude, Coding agents, Security, Privacy, Anthropic https://modelsatwork.news/article/cio-com-agent-autonomy-draw-the-line-at-the-action-not-the-agent Models at Worken 2026-11-06T10:10:00.000Z CIO.com: IT leaders draw agent autonomy lines per action, not per agent Agents, Human in the loop, Agent governance, Access control, Change management https://modelsatwork.news/article/anthropic-claude-ip-allowlist-admin-sign-out-api-leaked-key-controls Models at Worken 2026-11-06T02:00:00.000Z Anthropic adds IP allowlisting and a sign-out API to Claude admin controls Security, Access control, Claude, Anthropic, Vendor-published https://modelsatwork.news/article/box-content-retrieval-nvidia-nemotron-embed-agents-90-percent-fewer-tokens-techcrunch Models at Worken 2026-11-06T01:30:00.000Z Box builds Nvidia Nemotron into agent search, claims 90% fewer tokens RAG, Agent cost, Token cost, NVIDIA, Vendor-published https://modelsatwork.news/article/google-gemini-omni-1-2-flash-ga-15-second-video-30-percent-cheaper-7-cents-per-second Models at Worken 2026-11-05T23:58:00.000Z Google Gemini Omni 1.2 Flash: 15-second video at $0.07 a second Model release, Pricing, Google, Gemini API, Google Cloud https://modelsatwork.news/article/aws-web-answer-agentcore-cited-web-answers-10-dollars-per-1000-calls Models at Worken 2026-11-05T23:05:00.000Z AWS Web Answer for AgentCore: cited web answers at $10 per 1,000 calls Amazon Bedrock AgentCore, AWS, Agents, MCP, Pricing https://modelsatwork.news/article/gradium-phonon-65-mb-on-device-tts-apache-license-huggingface-vendor-published Models at Worken 2026-11-05T21:58:00.000Z Gradium's Phonon: 65 MB text-to-speech that runs on-device Open-weight models, Open source, Inference, Vendor-published, Self-hosting https://modelsatwork.news/article/openai-gpt-6-cyber-daybreak-red-948-linux-exploits-1-2-million-compute Models at Worken 2026-11-05T21:05:00.000Z OpenAI launches GPT-6 Cyber: 948 Linux exploits for $1.2M compute Model release, Security, OpenAI, Vulnerability disclosure, Agent safety https://modelsatwork.news/article/amazon-marketplace-a-plus-content-quality-vlm-ocr-catboost-94-percent-accuracy-vendor-published Models at Worken 2026-11-05T19:10:00.000Z Amazon: VLM plus OCR plus CatBoost scores listings at 94% accuracy Vendor-published, Case study, Evals, Amazon SageMaker AI, Monitoring https://modelsatwork.news/briefing/2026-11-07 Models at Worken 2026-11-07T11:30:00.000Z The briefing, 2026-11-07: AWS tests cutting off an AI agent: new requests stop in about 1 second, issued tokens live 5 minutes https://modelsatwork.news/briefing/2026-11-06 Models at Worken 2026-11-06T13:15:00.000Z The briefing, 2026-11-06: Users report Claude Code put their email addresses in web request headers; a fix is claimed from v2.1.317 https://modelsatwork.news/briefing/2026-11-06 Models at Worken 2026-11-06T11:30:00.000Z The briefing, 2026-11-06: Anthropic adds IP allowlisting and a sign-out-everywhere call to Claude, in beta