Hugging Face BlogFeatured6262The Hugging Face PyTorch team introduces updates to the MPS backend in PyTorch 2.15, adding the tensor torch.float8_e4m3fn and F.scaled_mm, with FP8 tensors occupying half the space of BF16.
Why it matters: The original text provides specific numbers for the MPS backend in FP8, MoE decoding, and memory usage, which can be used to judge the actual changes in running models locally on Mac.
Google Developers BlogCoral NPU has released a complete 32-bit architecture, including new matrix execution units, supporting up to 1 TOPS performance and 10mW average power consumption; launched the Coral Compiler v1.0 based on IREE/MLIR.
Mastra BlogMastra releases Mastra Launch service, providing enterprise-level access and dedicated support for teams already using the Mastra framework but not yet deploying agents into production within 90 days. The service includes full access to the Mastra Enterprise platform and weekly on-site collaboration support from Mastra engineers, aimed at helping teams complete production deployment within 90 days, otherwise only the implementation fee is charged.
Google Developers BlogArm leveraged the hierarchical graph visualization and plugin architecture of Google AI Edge Model Explorer to develop open-source adapters supporting formats such as TOSA, VGF, and PTE, helping developers more easily inspect edge model conversion processes.
Browser UseBrowser Harness achieves a score of 60.4 in matching evaluation using 2.1M median input tokens, matching Playwright MCP and Chrome DevTools MCP, becoming the browser control tool with the lowest cost per task at $0.042. Playwright MCP and Chrome DevTools MCP are faster but use more tokens.
Hugging Face BlogHugging Face has released TorchTPU, making TPU a native backend device for PyTorch. Transformers, Diffusers, and Kernels can now run directly on TPU. Users only need to select a model from the Hugging Face Hub and specify device_map="tpu".
Claude · YouTubeClaude has launched Claude Dashboards and Claude Motion, with the former supporting the real-time conversion of data platform or CRM data into dashboards in all paid plans, and the latter supporting the generation of animated explanatory videos from reports or charts in Team and Enterprise plans, both generated in code form without a video model, with content that can be modified and exported as MP4 or edited.
Transformers releases3535Transformers releases patch update v5.20.2, fixing an issue where input types were lost in the generate function, affecting multimodal models; it also addressed several other issues, including input dictionary type conversion, bfloat16 expert parallel gradients being zero, VLM input modal declaration for only reporting text, and host-to-device copy in eager_mask.
Meta EngineeringGenRecs Tokenization enhances conversion rates by 0.35% for advertisers by representing items like ads and products with a stable semantic vocabulary, combined with Meta's Generative Ad Model (GEM) and adaptive ranking model. This technology leverages the world knowledge of pre-trained large models to enable recommendation systems to understand preferences of new items or users with no historical interaction, accelerates learning through shared concepts, achieves cross-activity transfer, and improves performance as large model scale increases, forming a feedback loop for performance improvement.
Modal Blog7575NVIDIA has delivered its first Vera CPU server to Modal, an Arm-compatible CPU that, through emulation capabilities testing, has increased the throughput of the emulation capability sandbox by 2x, suitable for heterogeneous workloads with peak resource usage up to 10x the baseline, helping to reduce costs and accelerate reinforcement learning deployment.
SGLang releasesSGLang v0.5.22 released, new Model Type Cookbook, Clef, Clef Flash, PPLX-Decider v1/v1.1, EmbeddingGemma 2 added
Prime Intellect Blogprime-rl v0.10.0 has been released, defaulting to the new RL algorithm IPO to improve the stability of asynchronous RL and optimize performance. It supports new models such as Qwen3.8 Flash, and increases training throughput by 1.15 times (based on the torchtitan platform) through MoE optimization and custom kernels.
Inception Labs BlogMercury 2 decodes over 1000 tokens per second on standard NVIDIA GPUs, nearly 2 times faster than Gemini 3.1 Flash Lite in the query planning stage, 4.7 times faster than Claude Haiku 4.5, and 10 times faster than GPT-5 Mini. Each query step is faster than other models, allowing it to complete every step of the search process within the existing latency budget.
Inception Labs BlogMercury is launched on Azure AI Foundry, providing developers with the first commercially scalable diffusion large language model (dLLM), whose parallel generation capabilities based on diffusion architecture make its inference speed 10 times faster than similar autoregressive models.
Google Developers BlogLiteRT Tensor API is released, supporting developers to write and optimize models natively in C++ on edge devices, enabling cross-platform deployment (CPU/GPU/NPU) without script conversion. Developers can fully control the computational graph structure and directly load model weights, achieving zero-copy data transfer through shared accelerator buffers, while providing building blocks for AI coding agents.
Inception Labs BlogInception Labs has launched Mercury 2, increasing the number of free tokens from 10M to 100M. The new Inception API key includes 100M free tokens, and the free tier's request frequency has been increased 10 times. Mercury 2 performs better in real-time speech, search, AI coding sub-agents, and multi-agent workflows. The first token response time is reduced, end-to-end latency is lowered, and tool calls are more reliable.
Inception Labs BlogMercury Edit 2 is a new diffusion model (dLLM) launched by Inception Labs, used to predict the next step in code editing, with a 48% improvement in acceptance rate compared to the previous model, and more precise editing choices (improvement of 27%).
Hugging Face BlogTorchNeuron provides a new native backend for PyTorch on AWS Trainium, with native support for Transformers, Diffusers, and Kernels, making it unprecedentedly easy to run models on Trainium and share models on HF Hub.
Inception Labs BlogInception Labs launches Mercury Voice diffusion large language model, designed for voice agents, reasoning, calling tools, and following long system prompts, with a first response delay of less than 320 milliseconds, defeating GPT-6 Luna and Gemma 4 31B on proxies and voice benchmarks.
Amp News6565Assigning an issue to Amp in Linear will start a new Agent Session and generate a new orb. You can also start a session via @Amp. Amp will send its responses back to the same session and reply to messages received in the middle. The Agent Session will also appear in the Amp sidebar and can be viewed completely. Install the Amp agent to the Linear workspace and connect the Amp account to the Linear MCP to start using it.
Together AI BlogTogether AI has launched Post-Training services, supporting the construction of custom models through reinforcement learning and supervised fine-tuning without complex infrastructure. The service provides flexible training loop definition capabilities and optimizes the underlying system and infrastructure to improve model quality and training speed. It supports mainstream open-source models such as Muse Spark Open, GLM 5.3, Kimi K3, and Qwen 3.8-27B, running on dedicated GPU resources to ensure efficient and numerically consistent training.
Inception Labs Blog7575SearchBlox SearchAI now integrates Inception's Mercury dLLM, providing sub-second GenAI responses for enterprise workloads. This collaboration enables SearchBlox customers to achieve real-time search in e-commerce, customer service, knowledge management scenarios with 60-90% reduced reasoning costs and scalable speed.
Inception Labs BlogMercury 2 is released on Azure AI Foundry, the world's fastest reasoning language model, built to make production AI feel instant. The model combines Inception's breakthrough diffusion architecture with Azure's enterprise-grade infrastructure, providing developers with models that offer both exceptional speed and quality.
vLLM releasesvLLM v0.32.0 released, including 814 commits, contributions from 363 contributors, adding support for structured state Q&A via the POST /v1/systemone endpoint, new hf tools and parser support for general templates, low-precision KV cache optimization across all platforms, Model Runner V2 becomes the default for ROCm, startup and service performance improved by 43-85%
NVIDIA BlogNVIDIA RTX Spark laptops are now available, equipped with Blackwell RTX GPUs, up to 128GB unified memory, and the full NVIDIA AI platform, starting at $2399. They are launched by ASUS, Dell, HP, Lenovo, Microsoft, and MSI, bringing powerful local AI, content creation, and gaming capabilities to portable devices. NVIDIA held global launch events, including one in Berlin, where founder Jensen Huang made a surprise appearance.
Latent SpaceLatent Space has launched a new show 'High-End Tasters,' discussing topics such as Dev Day, the latest models, and assistants in the form of typical San Francisco lunchtime conversations.
Browser UseThe browser MCP server is divided into three types: local control, cloud execution, and task proxy, and needs to be selected according to the requirements. It is recommended to use Browser Harness for local control, Browser Harness CLI or Kernel for cloud execution, and Browser Use Cloud for task proxy. Browser Harness communicates directly with Chrome, without Playwright and extensions, but attention should be paid to security and performance issues.
LangChain Deep Agents releases1717LangChain Deep Agents released version 0.1.86, adding NVIDIA OpenShell as a sandbox provider, fixed multiple bugs including improved alias matching, accurate reporting of workspace rejection, replacing duplicate thread recovery errors, replacing chat input on recall prompts, displaying expiration prompts in the footer, providing actionable errors when Groq tools are limited, and comparing effective interpreter permissions, etc.
LangChain Deep Agents releases1010LangChain Deep Agents 0.7.28 released, fixes a bug where the `ContextHubBackend.glob` method did not handle the `path` parameter correctly.
LangChain Deep Agents releases3535LangChain Deep Agents 0.7.27 released, fixes a bug where child agent names were passed to child agents, and adds a feature to asynchronously report child agent status updates to the main agent.
OpenAI NewsItaú applies its AI-first strategy to customers and employees through the ia.i dialogue interface and the IARA governance platform, which is one of the most strictly regulated industries in Brazil.
O'Reilly Radar6565OpenAI's ChatGPT has added an Intelligent UI feature, capable of generating visual explanations and interactive tools; Anthropic's Claude Dashboards support natural language queries and visualizations of company data; Mistral has released the Large 4 model with trillion parameters for coding, agents, and multimodal tasks.
OpenAI NewsLocaliza uses ChatGPT Enterprise, OpenAI API, Codex, and workflows orchestrated by over 60 specialized AI models to build AI-native mobile experiences.
Ollama BlogReflection's first open-source weight model Beam has been fully launched this morning. Ollama is the first partner to provide this model for a limited time at no cost, and its Beam endpoint supports full zero-data retention (ZDR).
OpenAI News4545Stack Overflow for Agents achieved a 10-fold increase in intelligent agent creation through a ChatGPT plugin, developed by engineers using Codex in a few weeks.
Cursor ChangelogCursor Cloud Agent can build interactive Canvases and share them via links, Environments API has entered Beta program supporting programmatically creating and managing agent runtime environments, the large environment supports up to 300 repositories and loading speed is increased by 5 times.
Comfy Blog7575Krea 3 is now officially available in ComfyUI through Partner Nodes, a model designed for reference-based image generation and editing, supporting iterative changes through descriptions and fixed elements, aiming to speed up the process while maintaining image quality and reducing costs, allowing multiple edits to be included in a single session.
LangChain Deep Agents releases3535LangChain Deep Agents v0.7.26 released, asynchronous sub-agents can report waiting for human interaction status and display their running logs in the tracking of parent agents, while fixing the issue of lost tool result paths.
OpenAI News3838The GPT-Rosalind academic program is launched, offering sponsored access and tools to eligible academic laboratories to advance life science research.
