# Insight Daily — 2026-09-12 Independent, source-linked written edition. ## AI Daily 11 source-linked editorial stories. ### 1. Bedrock searches video scenes with Marengo 3.0 > AWS added TwelveLabs Marengo 3.0 multimodal embeddings to Bedrock Managed Knowledge Base. **Proof:** Results include segment start and end times, enabling applications to jump directly to the relevant moment in a video. **Impact:** The announcement does not establish error-free retrieval or specify regional eligibility and pricing. **Watch next:** Compare Marengo retrieval accuracy in a benchmark of known video moments and their returned timestamps. **Primary source:** https://aws.amazon.com/about-aws/whats-new/2026/09/amazon-bedrock-managed-knowledge-base-multimodal-embeddings-twelvelabs-marengo/ ### 2. Alibaba Token Plan adds metered agent tools > Alibaba says its personal Token Plan added Agent Harness tool benefits effective September 11 without changing existing prices or Credit allowances. **Proof:** 原有价格及Credit额度保持不变, Standard和Pro套餐新增Agent Harness工具权益与用量, 覆盖搜索、网页解析、图像生成、语音处理、代码执行等12项Agent开发常用能力。. **Impact:** Standard and Pro subscriptions add allowances for search, webpage parsing, code execution, speech and image tools through MCP. Subscribe on the Qianwen AI platform, obtain the Token Plan endpoint and API key, then configure a compatible tool. Included allowances vary by tool and plan. After they run out, additional calls are billed; video generation and several managed services are pay-as-you-go rather than included unlimited usage. **Watch next:** Check the Standard or Pro tool quota before a workflow; monthly tool allowances are separate from model Credits. **Primary source 1:** https://mp.weixin.qq.com/s/CXRjUVa-s26BFPLBVa4Guw **Primary source 2:** https://platform.qianwenai.com/pricing/token-plan ### 3. Claude plugin eval compares with and without your plugin > Claude Code added plugin behavior evaluation with a no-plugin baseline. **Proof:** Use evals to measure how reliably your plugin steers Claude to the right outcome, to catch regressions when you change the plugin or a new model ships, and to see what the plugin contributes compared with no plugin at all. **Impact:** Plugin authors need Claude Code 2.1.269 or later and a plugin manifest. Run claude plugin eval init at the plugin root, then claude plugin eval dot to test prompts against graders. By default each case runs three times with the plugin and three without. These and judge calls consume plan usage or API billing. A high score alone does not prove the plugin helped; compare the two arms. **Watch next:** Inspect failed graders and usage-limit errors before interpreting a score change as a regression. **Primary source 1:** https://code.claude.com/docs/en/plugin-evals **Primary source 2:** https://x.com/ClaudeDevs/status/2098500999656923145 ### 4. Cognition brings SWE-2 to Devin Desktop and CLI > Cognition released SWE-2, post-trained from Kimi K3, and made it available in Devin Desktop and CLI. **Proof:** SWE-2 is available starting today in Devin Desktop and CLI. We’re also rolling it out on Devin Web and Fusion. **Impact:** Engineers can use SWE-2 in those clients for coding work; Devin Web and Fusion are still described as rolling out. Cognition reports fewer detours and earlier implementation in its task benchmarks. Those are vendor evaluations under specific effort levels and task sets. Its pricing and benchmark comparisons are not guarantees that an individual project will be cheaper or correctly completed. **Watch next:** Compare accepted code and total cost in a SWE-2 benchmark using the same project and test suite. **Primary source:** https://cognition.com/blog/swe-2 ### 5. Kimi K2.8 Preview keeps the existing coding model ID > Kimi's September 11 update rolls K2.8 Preview into Kimi Code while keeping the kimi-for-coding model ID. **Proof:** K2.8 Preview 现已全量上线 Kimi Code, kimi-for-coding 直接升级、无需修改配置: 综合性能接近 K3, 支持 low / high / max 三档思考程度与最高 1M 上下文。详见 最新动态 。. **Impact:** The claim that performance approaches K3 is Kimi's assessment, not our independent benchmark. **Watch next:** Compare task quality and token cost in a K2.8 Preview benchmark that keeps reasoning effort constant. **Primary source 1:** https://www.kimi.com/code/docs/kimi-code/models.html **Primary source 2:** https://www.kimi.com/code/docs/kimi-code/whats-new.html ### 6. Sakana separates cost-focused Fugu Max from Ultra v2 > Sakana released Fugu Max and Fugu Ultra v2, two configurations of its model-orchestration architecture. **Proof:** Both models are available today via our standard OpenAI-compatible API. **Impact:** The benchmark comparisons are Sakana's claims, and SWEFish is its internal coding benchmark. They do not establish a guaranteed advantage on a customer's workload. **Watch next:** Compare Max and Ultra v2 on a workload benchmark using accepted outputs and total task cost. **Primary source:** https://sakana.ai/fugu-max-release ### 7. AWS compares model cost per successful outcome > AWS published an open benchmarking harness comparing practical model deployments by cost per correct answer and completed task. **Proof:** This is a comparison of practical deployment configurations, not a controlled estimate of intrinsic model capability. **Impact:** The results do not isolate intrinsic model capability. **Watch next:** Compare model cost per accepted deliverable in a benchmark that includes failed attempts and review work. **Primary source:** https://aws.amazon.com/blogs/machine-learning/beyond-the-price-per-token-choosing-the-right-openai-model-on-amazon-bedrock-for-your-workload/ ### 8. AgentCore separates working infrastructure from successful tasks > AWS published a dual-monitoring reference for an airline-reservation agent system. **Proof:** Infrastructure monitoring and agent effectiveness monitoring require different approaches. **Impact:** AgentCore Evaluations scores sampled interactions for helpfulness, correctness and goal completion. AWS DevOps Agent investigates infrastructure failures such as a missing model-invocation permission. Setup requires AWS account permissions and the Boto3 SDK. Model judges can be wrong, and asynchronous sampling can miss or detect a bad response only after delivery. This is an architecture example, not guaranteed autonomous incident resolution. **Watch next:** Compare evaluation judgments against known failures in an AgentCore benchmark; inspect missed goals despite healthy infrastructure. **Primary source:** https://aws.amazon.com/blogs/machine-learning/monitoring-production-agent-lifecycle-with-aws-devops-agent-and-agentcore-evaluations/ ### 9. AgentCore turns MCP responses into interactive booking cards > AWS published an AgentCore MCP Apps example that returns interactive HTML widgets inside compatible AI hosts. **Proof:** If the tool has an associated resource URI (for example, ui: //widget/unicorn-list ), the AI host initiates this phase. Tools without an associated widget, such as view_bookings and return_unicorn, return text-only content and skip this phase entirely. **Impact:** The sample uses an unauthenticated Gateway fronted by WAF; production access controls still need deliberate design. **Watch next:** Check the next AgentCore MCP Apps release note for rollout, availability, or quota changes. **Primary source:** https://aws.amazon.com/blogs/machine-learning/build-interactive-mcp-apps-using-amazon-bedrock-agentcore/ ### 10. NVIDIA NIM benchmark depends on workload and hardware > NVIDIA published a Nemotron 3 Ultra NIM serving configuration and performance comparison. **Proof:** The published curves are a starting point, not a promise that every application will see the same result. **Impact:** The reported throughput uplift combines caching, scheduling, precision and speculative decoding. It is not a general model-quality improvement, and the component gains cannot simply be added together. **Watch next:** Compare NIM configurations on a benchmark with the same hardware and per-user latency target. **Primary source:** https://developer.nvidia.com/blog/how-full-stack-nim-optimizations-deliver-2-5x-more-users-on-nemotron-3-ultra/ ### 11. SageMaker routes shared prompts to warm caches > AWS added prefix-aware routing for SageMaker real-time inference endpoints. **Proof:** You need at least two instances. With one instance, all requests go to the same place regardless of strategy. **Impact:** The existing Invoke API still works without changing model requests. At least two instances and serving-framework prefix caching are needed. Overloaded instances can spill requests elsewhere. AWS's reported latency improvement depends on its specific shared-prefix benchmark, not every prompt. **Watch next:** Compare cache-hit rate and latency in a SageMaker routing benchmark using consistent request serialization. **Primary source:** https://aws.amazon.com/blogs/machine-learning/reduce-llm-latency-with-prefix-aware-routing-on-amazon-sagemaker-inference/ ## Finance Daily 2 source-linked editorial stories. ### 1. Get ready to experience iPhone 18 Pro, the new Apple Watch lineup, and AirPods 5 > Get ready to experience iPhone 18 Pro, the new Apple Watch lineup, and AirPods 5. **Proof:** Apple Newsroom reports: Starting Saturday, September 12, customers can pre-order iPhone 18 Pro and iPhone 18 Pro Max. **Impact:** Pre-order availability lets customers commit to a purchase before delivery. Availability alone does not establish shipment volumes or realized sales. **Watch next:** Check the official delivery date and subsequent product sales disclosures to distinguish availability from uptake. **Primary source:** https://www.apple.com/newsroom/2026/09/get-ready-to-experience-iphone-18-pro-the-new-apple-watch-lineup-and-airpods-5/ ### 2. Salesforce Completes Acquisition of Fin > Salesforce Completes Acquisition of Fin. **Proof:** Salesforce Newsroom reports: Salesforce says the platform gives companies faster, more flexible ways to automate customer service and deliver measurable outcomes. That is the company’s claim; it does not establish measured customer savings. **Impact:** The business use is customer service automation. For companies using the platform, the financial question is whether that automation reduces the cost of handling customer requests while maintaining service quality. **Watch next:** After Salesforce Completes Acquisition of Fin, watch ACWI, EAFE, emerging markets, and official updates tied to those regions. **Primary source:** https://www.salesforce.com/news/press-releases/2026/09/10/salesforce-completes-acquisition-of-fin/ ### Verified source board The video admits every distinct verified story that fits the 600-second ceiling; this is the source-linked pool verified before selection. 1. **Christine Lagarde: Europe seen from Normandy** — European Central Bank Press Releases published an official update relevant to the market tape. ([source](https://www.ecb.europa.eu//press/key/date/2026/html/ecb.sp260912~fafa4b35b0.en.html)) - published/as-of: 2026-09-12T10:15:00+00:00; source: European Central Bank Press Releases; capture_sha256: 6481d714566148b6f2e09ab9cb85336fa24e804a9ecdc1c2fddc18c8ce6229c9 2. **From Wafer-Out to First Token: Codifying Supply Chain Expertise with Nemotron and Palantir Foundry** — NVIDIA has one of the largest and most complex supply chains in the world, and its performance is measured from wafer-out to first token. The interval is in two... ([source](https://developer.nvidia.com/blog/from-wafer-out-to-first-token-codifying-supply-chain-expertise-with-nemotron-and-palantir-foundry/)) - published/as-of: 2026-09-11T23:57:34+00:00; source: NVIDIA AI Blog; capture_sha256: 973da414f08c476196b361e22e06ad8706bb3867ec843328c099534da024a854 3. **8-K - Fly-E Group, Inc. (0001975940) (Filer)** — Filed: 2026-09-11 AccNo: 0001213900-26-099368 Size: 305 KB Item 5.02: Departure of Directors or Certain Officers; Election of Directors; Appointment of Certain Officers: Compensatory Arrangements of Certain Officers Item 9.01: Financial Statements and Exhibits ([source](https://www.sec.gov/Archives/edgar/data/1975940/000121390026099368/0001213900-26-099368-index.htm)) - published/as-of: 2026-09-11T21:29:31+00:00; source: SEC EDGAR Current Filings; capture_sha256: 437facb3d19592188007b033a6d0d77c808ab9847efc3167e3bac2e5f3d7bd5c 4. **8-K - SharonAI Holdings Inc. (0002068385) (Filer)** — Filed: 2026-09-11 AccNo: 0001493152-26-042434 Size: 343 KB Item 1.01: Entry into a Material Definitive Agreement Item 5.02: Departure of Directors or Certain Officers; Election of Directors; Appointment of Certain Officers: Compensatory Arrangements of Certain ([source](https://www.sec.gov/Archives/edgar/data/2068385/000149315226042434/0001493152-26-042434-index.htm)) - published/as-of: 2026-09-11T21:28:00+00:00; source: SEC EDGAR Current Filings; capture_sha256: 437facb3d19592188007b033a6d0d77c808ab9847efc3167e3bac2e5f3d7bd5c 5. **8-K - DeFi Development Corp. (0001805526) (Filer)** — Filed: 2026-09-11 AccNo: 0001805526-26-000104 Size: 1 MB Item 1.01: Entry into a Material Definitive Agreement Item 5.03: Amendments to Articles of Incorporation or Bylaws; Change in Fiscal Year Item 9.01: Financial Statements and Exhibits ([source](https://www.sec.gov/Archives/edgar/data/1805526/000180552626000104/0001805526-26-000104-index.htm)) - published/as-of: 2026-09-11T21:27:46+00:00; source: SEC EDGAR Current Filings; capture_sha256: 437facb3d19592188007b033a6d0d77c808ab9847efc3167e3bac2e5f3d7bd5c 6. **8-K - T3 Defense Inc. (0001787518) (Filer)** — Filed: 2026-09-11 AccNo: 0001185185-26-003953 Size: 339 KB Item 2.03: Creation of a Direct Financial Obligation or an Obligation under an Off-Balance Sheet Arrangement of a Registrant Item 9.01: Financial Statements and Exhibits ([source](https://www.sec.gov/Archives/edgar/data/1787518/000118518526003953/0001185185-26-003953-index.htm)) - published/as-of: 2026-09-11T21:25:26+00:00; source: SEC EDGAR Current Filings; capture_sha256: 437facb3d19592188007b033a6d0d77c808ab9847efc3167e3bac2e5f3d7bd5c 7. **8-K - KULR Technology Group, Inc. (0001662684) (Filer)** — Filed: 2026-09-11 AccNo: 0001104659-26-107130 Size: 188 KB Item 2.01: Completion of Acquisition or Disposition of Assets Item 5.02: Departure of Directors or Certain Officers; Election of Directors; Appointment of Certain Officers: Compensatory Arrangements of ([source](https://www.sec.gov/Archives/edgar/data/1662684/000110465926107130/0001104659-26-107130-index.htm)) - published/as-of: 2026-09-11T21:25:25+00:00; source: SEC EDGAR Current Filings; capture_sha256: 437facb3d19592188007b033a6d0d77c808ab9847efc3167e3bac2e5f3d7bd5c 8. **M 6.5 - 126 km NNE of Teluknaga, Indonesia** — M 6.5 - 126 km NNE of Teluknaga, Indonesia ([source](https://earthquake.usgs.gov/earthquakes/feed/v1.0/detail/us7000tgrk.geojson)) - published/as-of: 2026-09-11T21:23:55.907000+00:00; source: finance.usgs_earthquake; capture_sha256: 8361981950a617fe6cfa767aab43cd7cf19c09386f106ab249ba15a58356a36d 9. **Amazon EC2 X2idn instances are now available in Asia Pacific (Hong Kong)** — Memory-optimized Amazon Elastic Compute Cloud (Amazon EC2) X2idn instances are now available in Asia Pacific (Hong Kong) Region. These instances, powered by 3rd generation Intel Xeon Scalable Processors and built with AWS Nitro System, are designed for memory. ([source](https://aws.amazon.com/about-aws/whats-new/2026/09/ec2-x2idn-asia-pacific-hong-kong/)) - published/as-of: 2026-09-11T18:35:00+00:00; source: AWS What's New; capture_sha256: 5d7d7552acaecd22c0b2b7fbb60d9e83cf1f9f560e96616eef94e31b17d0fe5e 10. **Monitoring production agent lifecycle with AWS DevOps Agent and AgentCore Evaluations** — Multi-agent systems fail in ways traditional monitoring misses. This post presents a dual-layer approach to monitoring production agents: Amazon Bedrock AgentCore Evaluations for continuous quality scoring and AWS DevOps Agent for autonomous infrastructure inv ([source](https://aws.amazon.com/blogs/machine-learning/monitoring-production-agent-lifecycle-with-aws-devops-agent-and-agentcore-evaluations/)) - published/as-of: 2026-09-11T18:26:38+00:00; source: AWS Machine Learning Blog; capture_sha256: 1b8abce087c2cfd3e95c10fe2f8be797d5328a48ef55a5ded00f9c2e450f4017 11. **Amazon SageMaker HyperPod now supports model caching for faster inference autoscaling and reduced cold starts** — Amazon SageMaker HyperPod now supports model caching, an inference optimization that pre-loads model weights and container images onto cluster nodes so pods start in seconds instead of minutes. When running LLM inference at scale for workloads like chat assis. ([source](https://aws.amazon.com/about-aws/whats-new/2026/09/sgm-hyperpod-model-caching-inf/)) - published/as-of: 2026-09-11T18:25:00+00:00; source: AWS What's New; capture_sha256: 5d7d7552acaecd22c0b2b7fbb60d9e83cf1f9f560e96616eef94e31b17d0fe5e 12. **Beyond the price per token: Choosing the right OpenAI model on Amazon Bedrock for your workload** — Comparing models on dollars per million tokens misses what production workloads actually pay for: outcomes. This post shares an open-source benchmarking harness that measures cost per correct answer, agent trajectory cost, and rubric-graded deliverable quality ([source](https://aws.amazon.com/blogs/machine-learning/beyond-the-price-per-token-choosing-the-right-openai-model-on-amazon-bedrock-for-your-workload/)) - published/as-of: 2026-09-11T18:24:38+00:00; source: AWS Machine Learning Blog; capture_sha256: 1b8abce087c2cfd3e95c10fe2f8be797d5328a48ef55a5ded00f9c2e450f4017 ## GitHub Daily 3 source-linked editorial stories. ### 1. Immich > Immich has a verified tagged release, v3.2.0. **Proof:** GitHub Repo — pick #1; tagged release v3.2.0; recent activity docs(server): document the UTC timestamp format for timeBucket — https: //github.com/immich-app/immich. **Impact:** Immich: High performance self-hosted photo and video management solution. **Watch next:** README, latest release, and issue health. **Primary source 1:** https://github.com/immich-app/immich **Primary source 2:** https://github.com/trending?since=daily&spoken_language_code=en **Primary source 3:** https://api.github.com/repos/immich-app/immich **Primary source 4:** https://github.com/immich-app/immich/blob/main/README.md **Primary source 5:** https://github.com/immich-app/immich/releases/tag/v3.2.0 **Primary source 6:** https://github.com/immich-app/immich/commit/7abd625af305b82a4d37b578ffc535c3d7d302dc ### 2. Build Your Own X > Build Your Own X has no confirmed tagged release in the captured release index. **Proof:** GitHub Repo — pick #2; a releases index without a confirmed tagged release; recent activity Merge pull request reference 1844 from ARJ544/fix-ai-model-anchor Fix — https: //github.com/codecrafters-io/build-your-own-x. **Impact:** Build Your Own X: Master programming by recreating your favorite technologies from scratch. **Watch next:** README, latest release, and issue health. **Primary source 1:** https://github.com/codecrafters-io/build-your-own-x **Primary source 2:** https://github.com/trending?since=daily&spoken_language_code=en **Primary source 3:** https://api.github.com/repos/codecrafters-io/build-your-own-x **Primary source 4:** https://github.com/codecrafters-io/build-your-own-x/blob/master/README.md **Primary source 5:** https://github.com/codecrafters-io/build-your-own-x/releases **Primary source 6:** https://github.com/codecrafters-io/build-your-own-x/commit/aa17439b62f384511a5561ce308e9598b94d8989 ### 3. Armorpaint > Armorpaint has a verified tagged release, 1.0. **Proof:** GitHub Repo — pick #3; tagged release 1.0; recent activity paint: script tab improvements — https: //github.com/armory3d/armorpaint. **Impact:** Armorpaint: Graphics Creation Tools. **Watch next:** README, latest release, and issue health. **Primary source 1:** https://github.com/armory3d/armorpaint **Primary source 2:** https://github.com/trending?since=daily&spoken_language_code=en **Primary source 3:** https://api.github.com/repos/armory3d/armorpaint **Primary source 4:** https://github.com/armory3d/armorpaint/blob/main/readme.md **Primary source 5:** https://github.com/armory3d/armorpaint/releases/tag/26.09 **Primary source 6:** https://github.com/armory3d/armorpaint/commit/9dfa410af23bd06aef86eeb84d364a82afe9c41b --- For education and research only. Not investment advice.