https://modelsatwork.news/article/figma-stanford-motiongraphicsbench-32-models-svg-animation-gpt-6-astra-67-percent-verification-programs Models at Worken 2026-11-10T22:55:00.000Z Figma's MotionGraphicsBench: four of 32 models pass over half its SVG tasks Benchmarks, Evals, Research, Open-weight models, Reasoning effort https://modelsatwork.news/article/google-deepmind-gemini-robotics-2-5-mercedes-benz-kitting-pilot-96-percent-apptronik-apollo-2 Models at Worken 2026-11-10T21:58:00.000Z DeepMind's Gemini Robotics 2.5 hits 96% in a Mercedes-Benz kitting pilot Model release, Google, Robotics, Agent safety, Benchmarks https://modelsatwork.news/article/google-cloud-gemini-for-healthcare-preview-read-only-epic-oracle-health-model-routing Models at Worken 2026-11-10T21:25:00.000Z Google Cloud's Gemini for healthcare enters preview, read-only on EHRs Google Cloud, Healthcare, Agents, Access control, Compliance https://modelsatwork.news/article/ibm-cuga-harness-same-model-glm-5-3-flash-appworld-30-to-81-percent-vendor-published Models at Worken 2026-11-10T20:15:00.000Z IBM: same model, different harness moved AppWorld scores from 30% to 81% Agents, Evals, Agent governance, Open source, Vendor-published https://modelsatwork.news/article/aws-bedrock-tokenomics-data-lake-per-invocation-token-cost-vendor-published Models at Worken 2026-11-10T19:05:00.000Z AWS sample turns Bedrock invocation logs into a per-prompt cost lake Amazon Bedrock, Cost allocation, Observability, Token cost, Vendor-published https://modelsatwork.news/article/anthropic-claude-sheets-whiteboards-data-ga-enterprise-default-on-19-november Models at Worken 2026-11-10T18:20:00.000Z Anthropic adds Claude Sheets, Whiteboards; Claude Data on by default 19 Nov Anthropic, Claude, MCP, Access control, Agent governance https://modelsatwork.news/article/tcs-agentic-test-orchestrator-bedrock-agentcore-88-percent-effort-bfsi-vendor-published Models at Worken 2026-11-10T18:25:00.000Z TCS says its AgentCore test agents cut scenario effort by 88% Amazon Bedrock AgentCore, Multi-agent, Vendor-published, Financial services, Multi-tenancy https://modelsatwork.news/article/google-gemini-4-carbon-available-api-2-10-introductory-price-4-20-december Models at Worken 2026-11-10T17:10:00.000Z Google's Gemini 4 Carbon is out on the API at $2/$10 until 30 November Model release, Pricing, Google, Gemini, Model selection https://modelsatwork.news/article/google-tpu-8tx-8i-max-confidential-inference-million-token-context-vendor-published Models at Worken 2026-11-10T16:05:00.000Z Google adds TPU 8tx and 8i Max for confidential inference Model release, Google Cloud, Security, Inference, Vendor-published https://modelsatwork.news/article/scale-labs-remote-labor-index-claude-fable-5-2-41-percent-freelance-projects-manual-review Models at Worken 2026-11-10T14:58:00.000Z Scale Labs: Claude Fable 5.2 completes 41.3% of freelance projects Benchmarks, Evals, Claude, Agents, Research https://modelsatwork.news/article/anthropic-prompting-claude-fable-5-2-guide-effort-approvals-code-ownership-subagents Models at Worken 2026-11-10T13:57:00.000Z Anthropic's Fable 5.2 prompting guide: effort, approvals, code ownership Claude, Anthropic, Coding agents, Reasoning effort, Human in the loop https://modelsatwork.news/article/google-deepmind-alphaproof-code-signal-protocol-lean-verification Models at Worken 2026-11-10T13:10:00.000Z DeepMind's AlphaProof Code verifies Signal's core protocol code Security, Research, Coding agents, Gemini, Vendor-published https://modelsatwork.news/article/aws-bedrock-claude-fable-5-2-aws-review-data-retention-fallback-credit-retry-covered-model Models at Worken 2026-11-10T12:00:00.000Z Claude Fable 5.2 on Bedrock needs a data-retention setting before first call Amazon Bedrock, Data retention, Claude, Vendor-published, Procurement https://modelsatwork.news/article/ibm-palette-20b-open-weight-deck-builder-designer-coder-repair-harness-presentbench Models at Worken 2026-11-10T11:05:00.000Z IBM Research's Palette: a 20B open model that builds slide decks Open-weight models, Self-hosting, Evals, Research, Vendor-published https://modelsatwork.news/article/dharma-ai-field-extractor-2b-model-invoice-json-extraction-benchmark-l40s-cost Models at Worken 2026-11-10T10:00:00.000Z Dharma-AI: a 2B model for document field extraction beats frontier APIs Small models, Document extraction, Self-hosting, Cost, Vendor-published https://modelsatwork.news/article/zalando-agentic-identity-broker-oauth-delegation-mcp-servers-open-source-hn-objections Models at Worken 2026-11-10T01:56:00.000Z Zalando open-sources a broker for agent OAuth delegation Agents, Open source, Access control, Show HN, Agent security https://modelsatwork.news/article/aws-quick-skills-agents-publish-vs-share-skill-md-failure-steps-shared-agent-access Models at Worken 2026-11-10T01:25:00.000Z AWS Quick guide: treat shared agent skills like code Agents, AWS, Change management, Access control, Vendor-published https://modelsatwork.news/article/sagemaker-inference-gpu-partitioning-mig-h100-slice-vs-a100-cost Models at Worken 2026-11-10T00:50:00.000Z SageMaker adds GPU partitioning; an H100 slice beat a whole A100 Amazon SageMaker AI, GPU, Inference, Cost, AWS, Vendor-published https://modelsatwork.news/article/amazon-bedrock-managed-knowledge-base-single-shot-rag-maxagentiteration-0-to-5 Models at Worken 2026-11-09T23:05:00.000Z Amazon Bedrock Knowledge Base adds single-shot RAG and passage-only modes RAG, Amazon Bedrock, AWS, Cost, Vendor-published https://modelsatwork.news/article/artificial-analysis-claude-fable-5-2-index-65-275m-tokens-cost-per-task Models at Worken 2026-11-09T21:57:00.000Z Artificial Analysis ranks Claude Fable 5.2 first at 65, with 3.8x median tokens Benchmarks, Evals, Token cost, Claude, Anthropic https://modelsatwork.news/article/aws-pluggable-agentic-ai-framework-seven-layer-reference-architecture-agentcore-evals-gate Models at Worken 2026-11-09T21:10:00.000Z AWS's seven-layer reference architecture for production agents Amazon Bedrock AgentCore, Agents, Multi-agent, Agent governance, Vendor-published, AWS https://modelsatwork.news/article/anthropic-claude-fable-5-2-mythos-5-2-price-cut-8-dollars-fast-mode-vendor-benchmarks Models at Worken 2026-11-09T20:10:00.000Z Anthropic releases Claude Fable 5.2 and Mythos 5.2 at $8 per million tokens Model release, Pricing, Anthropic, Claude, Benchmarks https://modelsatwork.news/article/crossnokaye-aura-industrial-agents-bedrock-agentcore-verified-python-read-only-routines Models at Worken 2026-11-09T19:10:00.000Z CrossnoKaye's AURA agents run read-only plant checks on AgentCore Amazon Bedrock AgentCore, Agents, Agent safety, Human in the loop, Vendor-published, AWS https://modelsatwork.news/article/greenlight-financial-luna-90-percent-security-cases-resolved-bedrock-agentcore Models at Worken 2026-11-09T18:10:00.000Z Greenlight says its Luna agents resolve 90% of 4,000 monthly SOC cases Amazon Bedrock AgentCore, Multi-agent, Security, Human in the loop, Vendor-published, AWS https://modelsatwork.news/article/anthropic-claude-platform-notes-opus-4-8-fast-mode-removed-30-nov-ip-allowlist-message-threads Models at Worken 2026-11-09T17:20:00.000Z Anthropic ends Opus 4.8 fast mode on 30 Nov; adds IP allowlist beta Anthropic, Claude, API, Cost, Security https://modelsatwork.news/briefing/2026-11-10 Models at Worken 2026-11-10T11:30:00.000Z The briefing, 2026-11-10: Artificial Analysis ranks Claude Fable 5.2 first at 65, and says it used 3.8 times the median output tokens https://modelsatwork.news/briefing/2026-11-09 Models at Worken 2026-11-09T18:30:00.000Z The briefing, 2026-11-09: Anthropic says Opus 4.8 fast mode stops working on 30 November, and requests will fail rather than slow down