https://modelsatwork.news/article/worldwide-express-freight-billing-exceptions-three-bedrock-agents-75-percent-backlog-cut-vendor-published Models at Worken 2026-11-01T04:30:00.000Z Worldwide Express: three Bedrock agents cut billing backlog 75% Multi-agent, Human in the loop, Amazon Bedrock, AWS, Vendor-published https://modelsatwork.news/article/autobench-router-benchmark-no-llm-router-beats-always-calling-one-mid-priced-model Models at Worken 2026-11-01T00:00:00.000Z AutoBench: no LLM router beat one mid-priced model on cost and quality Routing, Evals, Model selection, Cost, Benchmarks https://modelsatwork.news/article/anthropic-text-watermark-four-more-claude-models-opus-4-6-sonnet-4-6-sampling-loophole-eu-ai-act Models at Worken 2026-10-31T23:00:00.000Z Anthropic widens Claude text watermark; two older models can dodge it Anthropic, Claude, Compliance, Regulation, Governance https://modelsatwork.news/article/aws-bedrock-prompt-caching-agents-gpt-6-astra-62-percent-lower-input-cost-stable-tools-prefix-vendor-published Models at Worken 2026-10-31T22:40:00.000Z AWS: prompt caching cut agent input cost 62% on Bedrock, in one test AWS, Amazon Bedrock, Agent cost, Token cost, Vendor-published https://modelsatwork.news/briefing/2026-10-31 Models at Worken 2026-11-01T00:30:00.000Z The briefing, 2026-10-31: AWS says prompt caching cut agent input cost about 62% on Bedrock, if tool order stays fixed