Together AI BlogTogether Link connects your team's existing tools to Together AI's top open-source models, reducing costs by over 50%. Supports Claude Code, Claude Desktop, Codex, OpenCode, and Pi, allowing switching without learning new tools, and can start using with a single command.
OpenAI Agents SDK releases77OpenAI Agents SDK released version v0.23.1, fixing bugs related to release test recovery and readiness checks, and passed the release readiness review. API version is 0.23.1, including 28 file changes (added 622 lines / deleted 42 lines), involving release automation, test fixtures, integration/examples tools, and release metadata.
GitHub Blog · AI & MLDeveloper workflows are changing due to AI, with skills beyond coding—such as guiding AI, evaluating outputs, communicating trade-offs, and making technical decisions—becoming increasingly important. Three key skills developers should strengthen are learning to guide AI rather than just using it, not trusting AI's first answer and evaluating it, and leveraging AI to solve broader problems.
Google ResearchGoogle Research announced a new generation of a federated learning system based on trusted execution environments (TEE), achieving fully verifiable and auditable data anonymization by executing encrypted training data within the TEE and publishing access policies to the public transparent log Rekor, Gboard has adopted this system and significantly improved computational speed.
NVIDIA BlogThe NVIDIA DGX Spark with 64GB unified memory, launched by Acer, ASUS, Dell, Gigabyte, HP, and MSI, supports developers, researchers, and AI enthusiasts to run models with up to 100 billion parameters locally without cloud dependency. Two devices can be clustered via NVIDIA Sync Cluster Assistant, merging memory to 128GB and boosting performance 1.7 times, priced from $4999.
OpenAI Agents SDK releases1515OpenAI Agents SDK v0.23.0 released, adding configurable limits to MCP list page, Docker removal protection for sandbox, and memory consolidation round settings, and introducing encrypted history scan budget options. Fixed issues with asynchronous cleanup on coroutine shutdown, CodeQL action grouping, and release source verification bugs.
NVIDIA BlogGPT-6 Astra Ultrafast version is available for OpenAI API and ChatGPT Work, Codex users, leveraging inference optimization using NVIDIA Blackwell architecture through OpenAI models, with a token generation speed 8 times faster than Astra Standard mode.
LangChain Deep Agents releases1212LangChain Deep Agents 0.1.80 released, fixing a bug that kept the session record position when opening a selector, and improving skill path handling and cache expiration notification features.
LangChain BlogLangChain has developed a model router for its open-source coding agent Open SWE, which dispatches requests to different models (GLM-5.3-Flash, GPT-5.6 Sol, GPT-6 Astra) based on task type and complexity, balancing cost, speed, and intelligence, reducing the median cost per thread by 64% without compromising quality.
Microsoft AI NewsMAI-Transcribe-2-Streaming provides low-latency real-time transcription for 60 languages and supports automatic language detection, achieving the top accuracy in both final and partial transcription on Artificial Analysis.
Microsoft Official Blog4040Ryan Roslansky decided to leave Microsoft and LinkedIn after nearly 18 years, will serve as a consultant before the end of the year and support the transition; Nadella and Roslansky emphasized the rapid progress of Copilot in building new operating systems, Office is fully embedded in Copilot and vice versa; LinkedIn will continue to focus on its core vision of creating economic opportunities for its 1.3 billion members worldwide.
Amazon ScienceAmazon Science has released Graph Center Agent technology, which combines a continuous synchronized graph (digital twin) representing the network with device attributes and connections, with cascade graph algorithms (decomposition, aggregation, tracing) and agent orchestration to achieve minute-level root cause analysis of complex network failures, achieving 95% accuracy in a demonstration at NTT DOCOMO's MWC.
Claude.dev BlogClaude Code 2.1.287 or higher supports mods, allowing users to request Claude to generate JavaScript or TypeScript files to extend its functionality. Mods can observe sessions, modify behavior, or draw custom UIs without learning APIs. For example, mods can add real-time context window status displays (like Token Weather), intercept commands, or build custom interactive interfaces, these hooks are implemented through plugins and override all events in a session.
Anthropic ResearchFeatured7878Anthropic researcher Matt reflected on the experience of using Claude with the open-source harness BootLoops for research, advancing 36 papers in 18 fields with 19 collaborators in three months, selecting from about 400 candidate problems.
Why it matters: The author reflects on first-hand interdisciplinary practice, exploring the real boundaries of using agents for research, and providing reusable open-source harnesses and collaboration methods.
Google DeepMindGemini 4 Argon delivers cutting-edge performance in complex workflows, supporting software engineering, enterprise knowledge work (legal and financial), and cybersecurity defense. The model's token output limit is extended to the industry-leading 1M, and it leads in benchmarks such as DeepSWE v1.1 (77.9%), Vals Index (financial, coding, legal, tax), and AutomationBench (51.3%).
LangChain Deep Agents releases3232LangChain Deep Agents 0.7.21 released, adding `FilesystemMiddleware` for blob storage functionality in the backend. Updated urllib3 dependency to 2.8.0 and increased the minimum version requirement for dependencies.
Microsoft ResearchThe machine learning system developed by Microsoft Research can predict the spatial weather risks of 66,935 substations in the United States, combining solar wind observations, auroral electrojet (AE) indices, and disturbed storm time (Dst) index forecasts.
Google DevelopersFeatured2222Google's October 1 Developer Knowledge release notes add knowledge.workspace.google.com to the corpus. Earlier September updates made the developer-knowledge CLI generally available and introduced an agent skill that selects an MCP tool or falls back to the REST API. The API provides documentation search, document retrieval and grounded answers.
Why it matters: Editorial takeaway: documentation retrieval can be part of an agent's toolchain instead of relying entirely on memorized APIs. Check supported domains and access requirements; this is a corpus update, not the API's original launch.
Anthropic NewsroomFeaturedAnthropic says Barclays is expanding Claude across software development, legacy modernization and operations. Its knowledge assistant has been adopted by more than 16,000 colleagues and handled over one million searches; a separate email workflow processes roughly 120,000 emails a day. The announcement describes governance and human oversight alongside these deployments.
Why it matters: Editorial takeaway: this is an adoption example, rather than a new model release. It illustrates where enterprise AI is being applied; the reported outcomes come from the partnership announcement.
Google DeepMindGoogle DeepMind has launched SynthID Bio watermark technology, embedding imperceptible identifiers into biological code to ensure verification on both digital models and synthetic proteins while preserving the biological function of the proteins.
Anthropic ResearchFeatured7171Anthropic has released a Robot Exposure Index study, using Claude to evaluate about 19,000 tasks in O*NET, finding that existing robots can complete 74% of physical tasks in the U.S. and account for 34% of all work time, but most are limited to controlled environments.
Why it matters: Anthropic used Claude to rate O*NET tasks item by item, quantifying how much work robots can currently cover, and provided constraints on cost and regulation.
Google Developers BlogSparse VideoGen (SVG) dynamically routes attention heads to spatial or temporal masks by identifying the spatiotemporal structures of different attention heads in video diffusion models, converting theoretical sparsity into actual end-to-end speedup. On TPU v6e, SVG retains about 38.87% of query-key pairs, achieving 3.5x inference acceleration while maintaining output quality at a given sparsity level. The scheme employs custom JAX and Pallas Splash attention kernels to convert theoretical sparsity into physical sparsity on hardware.
Microsoft ResearchMicrosoft Research has released Quine, an AI research platform that combines a multi-modal biological world model with an interactive connection system, used to integrate models, scientific tools, literature, and researchers, and has been used to predict compounds driving tumor state changes and validate candidates; the system is designed for research, not for clinical or medical use, and its output may be incomplete or require verification by researchers, with future plans to expand access through Microsoft Discovery.
Microsoft Official BlogMicrosoft launched Fabric IQ, Azure Databases, and other data innovations at the European Microsoft Fabric + SQL Community Conference in Barcelona, integrating unique business contexts into Microsoft Copilot.
Ollama BlogOllama 0.35 version adds support for decision models based on the TypeSafe Jev API through the new /v1/systemone endpoint, allowing local operation for low latency, typed rapid decision-making.
Anthropic ResearchZhipu AI's Released GLM-5.3 Model Successfully Developed End-to-End Exploitation Tools 50 Times in ExploitBench Tests; 4% of Tasks Achieved Complete Control Flow Hijacking in Internal Binary Exploitation Tests, but the Model Lacks Effective Security Protection Measures; Simulation Tests Show Attackers Can Intercept and Evade 64%-100%, Far Exceeding the Defense Effectiveness of Protected Models.
OpenAI Deployment SafetyGPT-6.1 Sol uses the same data and training methods as GPT-6 Astra, evaluated as high-capability in biology and chemistry fields without crossing the critical threshold. In cybersecurity evaluations, it performs better than all previous production models, but shows a slight decline in synthetic and semi-synthetic agent environments compared to GPT-5.6 Sol. Model rejection remains a layer of security boundary.
Microsoft ResearchMicrosoft Research Asia – Singapore (MSRA – Singapore) officially opened on July 24, 2025, becoming Microsoft's first research lab in Southeast Asia
Google AI Blog5050The independent film 'The Gifted' won the Future Vision XPRIZE, with a prize of 1 million USD and 2.5 million USD in film production funding. The film tells the story of an 11-year-old boy who uses code to recreate his mother's voice to care for the community, standing out from 2500 global works. Google, through the 100 ZEROS program, partnered with Range Media Partners to support the film's release.
Mistral AIMistral AI opens a German center in Munich to advance industrial AI in Europe's largest economy. Germany is the EU's largest industrial economy, with its companies accumulating multi-generational proprietary data and process knowledge, covering automotive, energy, aerospace, and advanced manufacturing sectors.
Claude.dev BlogClaude Sonnet 5.5 is the second model in the Claude 5.5 family after Opus 5.5, more intelligent, efficient, and 30% faster than Sonnet 5, but the price per token remains unchanged, as the cost is reduced due to a maximum of 30% reduction in token consumption for typical tasks.
Claude app release notesAnthropic releases Claude Sonnet 5.5, the second model in the Claude 5.5 family, as a faster and more affordable supplement to Claude Opus 5.5.
AnthropicFeatured5555Anthropic introduced Sonnet 5.5 as a faster complement to Opus 5.5 for scoped tasks, bug fixes and document creation. Its release announcement reports performance and efficiency improvements over Sonnet 5.
Why it matters: Editorial takeaway: compare models against your own coding and document tasks. The release emphasizes everyday work, while distinguishing more complex, open-ended workloads.
Xiaomi MiMo · Hugging Face new modelsXiaomi MiMo team releases MiMo-V2.6-Flash-MOPD model, integrating multiple domain expert teachers to enhance reliable training and validation capabilities in long-term game development, research, and embodied intelligence, and effectively alleviates the issue of repetitive tool calls.
Xiaomi MiMo · Hugging Face new modelsXiaomiMiMo/MiMo-V2.6-Pro-MOPD integrates multiple domain expert teachers, expanding to long-term game development, scientific research, and other fields where reliable training and validation are difficult.
GitHub Blog · AI & MLThe canvas in GitHub Copilot is a customizable interface that can be shared between users and agents, supporting various forms like Kanban, issue filtering, and release lists. It can be automatically generated by describing the requirements via the /create-canvas skill, and both parties can collaborate in real-time to modify the status.
Amazon ScienceReactor and Amazon Neuron Science jointly optimize the Rolling Forcing model to achieve real-time video generation on Trainium, which combines the sequence prediction of large language models with the iterative optimization of diffusion models, running at 16 FPS standard video frame rate, generating 30-second high-quality videos through Rolling Forcing technology, and achieving low latency, high-efficiency real-time video generation on Trainium.
.png&mode=card&exp=1794873600&sig=3d0151be74acba48)
