https://modelsatwork.news/article/worldwide-express-freight-billing-exceptions-three-agents-bedrock-10000-weekly-vendor-published Models at Worken 2026-10-30T17:58:00.000Z Worldwide Express: 3 agents close 10,000 freight billing exceptions a week Vendor-published, Multi-agent, Human in the loop, Amazon Bedrock, Case study https://modelsatwork.news/article/anthropic-claude-tag-api-beta-slack-channel-settings-as-code-14-endpoints-claude-enterprise Models at Worken 2026-10-30T17:40:00.000Z Anthropic's Claude Tag API lets IT manage Claude in Slack from code Anthropic, Claude, Access control, Agent governance, Enterprise adoption https://modelsatwork.news/article/anthropic-usage-and-cost-analytics-api-beta-spend-by-actor-api-key-workspace-4-hour-lag Models at Worken 2026-10-30T16:15:00.000Z Anthropic's new Usage and Cost API reports Claude spend by user and key Anthropic, Cost allocation, Cost governance, API, Monitoring https://modelsatwork.news/article/autobench-router-benchmark-no-commercial-router-beat-always-luna Models at Worken 2026-10-30T15:20:00.000Z AutoBench: no commercial LLM router beat a single mid-tier model on cost and quality Model selection, Cost, Evals, Benchmarks, Agent cost https://modelsatwork.news/article/doordash-several-hundred-orders-from-third-party-ai-agents-vs-3-million-daily-deliveries-techcrunch Models at Worken 2026-10-30T14:30:00.000Z DoorDash: only 'several hundred' orders have come from AI agents Agents, Personal agents, Enterprise adoption, MCP, Case study https://modelsatwork.news/article/epoch-ai-glm-5-3-firefox-patch-gap-exploits-1000-dollars-per-release Models at Worken 2026-10-30T13:35:00.000Z Epoch AI: open-weight GLM 5.3 exploited six Firefox releases for ~$1,000 each Security, Open-weight models, Evals, Research, Agent safety https://modelsatwork.news/article/anthropic-study-50-analysts-followed-claude-code-coin-flip-lean-86-percent-judgment-calls Models at Worken 2026-10-30T12:05:00.000Z Anthropic: analysts followed a coin-flip AI lean 86% of the time Research, Human oversight, Agents, Anthropic, Agent safety https://modelsatwork.news/article/aws-batch-quota-management-sagemaker-training-gpu-lend-borrow-preemption-vendor-published Models at Worken 2026-10-30T11:05:00.000Z AWS Batch adds GPU quota shares that teams can lend and borrow Vendor-published, AWS, Amazon SageMaker AI, Cost governance, Inference https://modelsatwork.news/article/openai-api-credits-chatgpt-pro-business-subscribers-500-dollars-month-expire-no-rollover-vendor-published Models at Worken 2026-10-30T09:58:00.000Z OpenAI gives ChatGPT Pro and Business plans up to $500 a month in API credits OpenAI, Pricing, Cost, Vendor-published, Procurement https://modelsatwork.news/article/epsilon-nova-act-ui-testing-platform-service-playwright-vendor-published Models at Worken 2026-10-29T19:58:00.000Z Epsilon runs UI tests in plain English on Nova Act, beside Playwright Amazon Nova Act, Amazon Bedrock AgentCore, Browser agents, Agent evaluation, Vendor-published https://modelsatwork.news/article/aws-kiro-php-to-java-migration-engine-parity-harness-first-4-of-37-models-vendor-published Models at Worken 2026-10-29T18:56:00.000Z AWS: build the parity test before the Java in a Kiro PHP migration Vendor-published, Coding agents, AWS, Verification, Developer tools https://modelsatwork.news/article/spawn-agents-ship-hourly-30x-payroll-token-spend-simon-willison Models at Worken 2026-10-29T18:00:00.000Z Spawn ships hourly with agents, spends 30x payroll on tokens Agents, Coding agents, Cost, Human in the loop, Developer tools https://modelsatwork.news/article/moengage-400-apis-agent-tools-agentcore-gateway-cedar-policies-vendor-published Models at Worken 2026-10-29T17:10:00.000Z MoEngage exposes 400 private APIs to 10 agents through AgentCore Gateway Vendor-published, MCP, Access control, Amazon Bedrock AgentCore, Agents https://modelsatwork.news/article/artificial-analysis-gemini-4-carbon-index-v5-59-cost-per-task-7-76-hallucination-rate-11-percent-secondary-report Models at Worken 2026-10-29T16:20:00.000Z Artificial Analysis: Gemini 4 Carbon tops its index at $7.76 per task Benchmarks, Evals, Model selection, Gemini, Cost https://modelsatwork.news/article/mistral-large-4-open-weights-released-modified-mit-licence-coding-cyber-vendor-benchmarks Models at Worken 2026-10-29T15:15:00.000Z Mistral releases Large 4 weights: 1T-parameter MoE, 1M context Mistral, Model release, Open-weight models, Self-hosting, Coding agents https://modelsatwork.news/article/cio-ai-vendor-failure-exit-plan-export-data-revalidation-sorokin-shanker Models at Worken 2026-10-29T14:30:00.000Z CIO: plan your AI vendor's exit before you sign Procurement, Vendor lock-in, Evals, Enterprise adoption https://modelsatwork.news/article/anthropic-claude-for-legal-artifacts-cite-check-playbook-deviation-board-60-connectors-vendor-published Models at Worken 2026-10-29T11:05:00.000Z Anthropic adds legal artifacts to Claude Enterprise, with cite checks Vendor-published, Anthropic, Claude, Human in the loop, Verification https://modelsatwork.news/article/simon-willison-vibe-coded-games-opus-5-5-single-prompt-playtest-bugs Models at Worken 2026-10-29T09:58:00.000Z Simon Willison: Claude Opus 5.5 now builds games that play well Coding agents, Claude, Model selection, Open source, Prototyping https://modelsatwork.news/article/anthropic-claude-enterprise-smart-reports-pooled-group-budgets-spend-per-workstream-vendor-published Models at Worken 2026-10-28T21:58:00.000Z Claude Enterprise adds smart reports and pooled team budgets Anthropic, Claude, Cost governance, Enterprise adoption, Vendor-published https://modelsatwork.news/article/aws-nova-act-qa-studio-part-3-test-health-score-flaky-penalty-team-roles-vendor-published Models at Worken 2026-10-28T20:58:00.000Z AWS QA Studio: a 0-100 test health score that penalises flaky tests Amazon Nova Act, AWS, Agent evaluation, Access control, Vendor-published https://modelsatwork.news/article/ai2-novelty-judge-bench-prompt-wording-swings-llm-judge-accuracy-93-to-41-percent Models at Worken 2026-10-28T19:58:00.000Z Ai2: rewording one prompt dropped an LLM judge from 93.5% to 40.9% Evals, LLM-as-judge, Benchmarks, Research, Cost https://modelsatwork.news/article/aws-agentcore-enterprise-agent-platform-blueprint-workstream-cells-eval-gate-denial-tests-vendor-published Models at Worken 2026-10-28T18:58:00.000Z AWS blueprint: an agent platform where failed denial tests fail the build Amazon Bedrock AgentCore, AWS, Governance, Evals, Vendor-published https://modelsatwork.news/article/google-gemini-4-carbon-cyber-defenders-first-api-price-same-as-argon-vendor-published Models at Worken 2026-10-28T18:05:00.000Z Google announces Gemini 4 Carbon, open first to cyber defenders Model release, Gemini, Google, Security, Pricing https://modelsatwork.news/briefing/2026-10-30 Models at Worken 2026-10-30T15:30:00.000Z The briefing, 2026-10-30: Anthropic says 50 analysts took their coding agent's random lean on 86% of judgment calls https://modelsatwork.news/briefing/2026-10-29 Models at Worken 2026-10-29T10:30:00.000Z The briefing, 2026-10-29: Anthropic says Claude Enterprise now shows spend per team workstream, with pooled monthly budgets https://modelsatwork.news/briefing/2026-10-30 Models at Worken 2026-10-30T10:30:00.000Z The briefing, 2026-10-30: OpenAI gives ChatGPT Pro and Business plans up to $500 a month in API credits that expire each cycle