Amazon has enabled usage-based software metering for bidirectional streaming models on SageMaker AI. Previously, software charges for long-running streaming sessions were difficult to capture, as they could not be reconciled with simple HTTP response headers. Model providers can now report usage incrementally via a dedicated metadata WebSocket connection.
MiniMax has launched MiniMax-H3-Image-PE, a fine-tuned version of the Qwen3.8-27B model designed to transform brief text prompts into highly detailed descriptions for image generation. The release includes specific workflows for text-to-image expansion and early-stage support for image editing prompts.
NVIDIA has released an open-source tool, NVIDIA Snapshot, for Kubernetes-based GPU workloads. By checkpointing fully initialized models and their states, the system allows inference replicas to resume in seconds rather than minutes, effectively eliminating repetitive initialization steps.
NVIDIA has contributed Agent Substrate to the CNCF Sandbox to solve the inefficiency of running AI agents on Kubernetes. The platform allows multiplexing many agent sessions onto a small pool of warm worker pods by checkpointing memory and filesystem states during idle periods.
Amazon introduced a seven-layer reference architecture to move AI agent applications from initial prototypes to production-ready systems. The framework emphasizes a modular, pluggable design using stable contracts between layers such as data, inference, orchestration, and security.
Google Cloud has launched dynamic slicing for TPUs on GKE, allowing developers to decouple logical hardware configurations from physical nodepools. This update eliminates the need to manually recreate nodepools when workload requirements change, enabling faster scaling and improved fault tolerance.
NVIDIA has introduced deployment recipes for four prominent Korean open models, including A.X K2 and Solar Open 2, within the Dynamo framework. These recipes are specifically optimized for multi-node NVIDIA B200 GPU clusters, designed to handle large-scale inference efficiently.
Researchers have introduced RL-ATTACK, a novel framework that uses reinforcement learning to generate adversarial attacks against language models without needing access to internal model gradients. By learning from the target model's responses, this method achieves superior query efficiency and successfully bypasses safety filters by forcing unintended outputs.
NVIDIA has introduced new optimization tools in TensorRT LLM VisualGen for the FLUX 3 Image Klein model. By utilizing NVFP4 quantization, kernel fusion, sparse attention, and step caching, developers can achieve up to 3.46x faster text-to-image generation on NVIDIA Blackwell hardware.
Amazon SageMaker Inference has introduced support for NVIDIA Multi-Instance GPU (MIG), allowing you to partition a single physical GPU into multiple independent units. This hardware-level division provides dedicated compute and memory to each partition, ensuring predictable latency and strong isolation between models hosted on the same accelerator.
Taxbit has deployed a secure, role-aware AI assistant using Amazon Bedrock AgentCore to automate product documentation support. The agent uses RAG to ground answers in verified content while strictly enforcing PII masking, regional data residency, and user-role permissions.
Google has launched Gemini Omni 1.2 Flash, a model update focused on video production. It extends native video generation to 15 seconds, improves object and character consistency, and introduces multi-turn conversational editing.
NVIDIA has introduced five co-design pillars for its Vera Rubin NVL72 platform, aimed at maximizing throughput in agentic AI workloads. By combining new precision formats like NVFP4, wide expert parallelism, and the Dynamo distributed serving layer, the system treats memory as a unified, tiered resource across GPUs, host CPUs, and specialized storage.
Gradium has released Phonon, a suite of compact (65-75 MB) text-to-speech models designed to run entirely on edge devices. It supports five languages, operates offline, and performs competitively against models significantly larger in size.
NVIDIA has launched CUDA 13.5, reaching general availability for its new Vera Rubin GPU architecture. This update introduces significant enhancements to multi-process service (MPS) resource scheduling, locality domain partitioning, and deeper integration with SIMT libraries via CUDA Tile.
A new method called FlagRot optimizes key-value (KV) cache quantization for long-context models. By selectively rotating key segments based on attention error rather than applying global rotation, it reduces accuracy loss in 2-bit or 4-bit caches.
New research shows that reasoning history can be rewritten into compact state records before new tool results are introduced. This method effectively reduces future reasoning tokens and total agent spending by eliminating redundant thoughts without requiring model retraining.
Researchers have developed a method to calculate the theoretical upper bound for Multi-Draft Speculative Decoding (MDSD) efficiency by solving the dual of the optimal transport problem. The study reveals that current verification algorithms often fall short of this theoretical limit, and that sampling without replacement significantly improves acceptance rates compared to sampling with replacement.
NVIDIA has introduced Cosmos, a platform offering world foundation models and development tools specifically designed for physical AI applications, such as robotics and autonomous vehicles. The framework includes modular training recipes and NIM microservices for GPU-accelerated inference, enabling the generation of synchronized video and LiDAR observations.
Amazon has published a workflow for using Sequence-Level Knowledge Distillation (SeqKD) to fine-tune smaller models on trajectories generated by frontier models. By using a 'teacher' model to generate successful tool-use trajectories and filtering them with a programmatic scorer, teams can train compact 'student' models that perform complex reasoning at a fraction of the cost.
NVIDIA has released Workload Map, a tool designed to provide a common way for Kubernetes schedulers and platforms to interpret diverse distributed AI workloads. By using declarative YAML definitions, it eliminates the need for developers to build custom parsing adapters for every new framework like JobSet, Ray, or KServe.
Apple researchers have unveiled an online training infrastructure for mobile UI agents that utilizes a two-part architecture to split navigational logic from visual interpretation. By employing step-level feedback provided by an LLM-as-a-Judge and a novel value estimation technique known as GiGPPO, these agents demonstrate significantly better adaptability than static models trained on pre-recorded sequences.
Recent research shows that many attention heads in large language models are redundant and specifically contribute to verbatim memorisation of training data. By selectively pruning these heads at inference time, researchers achieved a 91.5% reduction in memorisation events while simultaneously improving performance on standard benchmarks.
NVIDIA has released AI Cluster Runtime (AICR) to package Dynamo inference infrastructure into validated recipes. This update simplifies deployment by pinning versions and removing mandatory external dependencies like NATS, while introducing new KV-aware routing paths for Kubernetes.