Audible has developed a controllable text-to-speech pipeline that uses LLMs to profile characters and attribute dialogue with up to 97% accuracy. By providing 'stage directions' as natural-language captions, the system ensures distinct, consistent character voices over long-form content.
AWS published a walkthrough for the Adaptive Agentic Intelligence Framework (AAIF), built on Strands Agents, which routes simple requests to small local models (via Ollama) and only sends complex requests to cloud models through Amazon Bedrock. A lightweight 1B-parameter model (llama3.2:1b) acts as an orchestrator deciding where each request should go, with an example IoT monitoring agent that uses local telemetry retrieval, local vision analysis, and cloud-based expert insights. AWS states that in the hybrid setup, roughly 80 percent of traffic can be handled locally without hitting a paid API, and the architecture can run fully on edge hardware (e.g. Raspberry Pi with 8GB memory), fully in the cloud, or in a hybrid mode.
Microsoft reports that Storm-0940, a China-state aligned threat actor active since 2021, began using the open-source agentic AI penetration testing framework ARTEX in September 2026 to automate attacks against hundreds of organizations worldwide. Microsoft assesses the actor was behind the recently disclosed ConnectWise supply chain compromise, in which ARTEX-driven password spraying and OAuth application abuse reached over 2,250 downstream tenants at up to 2,900 sign-in attempts per minute. In several cases the actor obtained privileged cloud access in under 3 hours, and follow-on activity included credential and privilege changes aimed at persistence.
Cohere launched Rerank 5, its newest enterprise reranking models, in Pro and Fast tiers. Rerank 5 Pro leads major public benchmarks including BRIGHT, ViDoRe V3, BEIR and nanoMTEB, while Rerank 5 Fast offers near-leading quality at lower cost and higher throughput. The release introduces Context Focus, which returns the most relevant text spans within documents rather than whole documents, and calibrated relevance scores that read as probabilities across models and domains.
IBM Research released vLLM-Hook MIA, a plugin for the open-source vLLM inference engine that lets developers observe and intervene in a model's internal states at runtime — for example to decline malicious prompts, flag hallucinated facts, or adjust tone — without disabling vLLM's fast graph-replay execution mode. According to IBM's technical report, MIA is up to 20x faster than the earlier vLLM-Hook and up to 14x faster than methods that insert code directly into the model. In tests, steering an 8B Llama or 1.5B Qwen model added only about 2% latency, and a 70B Llama model added about 6%.
NVIDIA has added three new open, commercially licensed components to its Nemotron Retriever lineup: Embed models (8B and 1B, in BF16 and NVFP4) for searching text and images, a Rerank model (8B) that reorders retrieved evidence by relevance, and a compact 631M Parse model that converts document images into structured text. The goal is to feed AI agents with better, more relevant evidence so they need fewer search rounds and fewer inference tokens to answer a question. NVIDIA reports the new Embed and Rerank models top the ViDoRe v3.1 leaderboards for dense retrieval and reranking respectively.
AI research company i14 partnered with AWS to port its Laplace 3B model, a 24-layer architecture built on Gated Linear Attention (GLA) instead of standard transformer attention, onto AWS Trainium2 using native PyTorch (TorchNeuron Native). Rather than a KV cache that expands with every token, the model keeps a constant 29.5 MB recurrent state per sequence. On a single trn2.48xlarge instance, the team reached about 120,000 training tokens per second and served 11,544 output tokens per second across 1,280 simultaneous sequences through an unmodified vLLM engine.
Google Cloud has made Anthropic's Claude Opus 5.5 and Claude Sonnet 5.5 selectable inside its Antigravity agentic coding platform, alongside Gemini 3.8 Flash and Gemini 4 Argon. Developers can switch between model families mid-conversation across Antigravity 2.0, the CLI, and IDE extensions for VS Code, JetBrains and Visual Studio, all under one enterprise billing and security boundary. Google also describes two engineering patterns: using a second model family to independently review code written by the first ('LLM-as-a-judge'), and assigning fast Gemini Flash subagents to handle high-volume mechanical tasks while a stronger model handles deep reasoning.
Allen Institute for AI (Ai2) released FairStress, a benchmark that tests demographic fairness in LLMs not just by their final decisions but by how easily those decisions change under persuasive pressure. Testing open and frontier models including Olmo, Llama, Qwen, Gemma, Claude, and GPT found that models often abandon correct, well-supported decisions more readily when doing so favors a minority-coded candidate — a pattern the authors call 'anti-stereotypical over-correction.' The benchmark covers 300 scenarios across 29 domains (hiring, lending, healthcare, education, housing, criminal justice), expanded via demographic and persuasion variations into over 13 million test cases, plus a smaller 50K-item 'FairStress-Core' subset for faster evaluation.
Google introduced TorchTPU, a native PyTorch backend for Cloud TPUs (torch.device("tpu")), and co-developed a fork of Meta's TorchTitan trainer adapted for TPU training. Previously, running PyTorch training on TPUs meant rewriting code in JAX or relying on the older torch_xla path; now only minimal device-initialization changes are needed. The fork provides tested recipes for pre-training large dense models (Llama 3 70B/405B, Qwen 3 8B/32B), Mixture-of-Experts models (Qwen 3 235B), and diffusion transformers (FLUX.1 12B), plus a 'Graph Trainer' that captures an entire training step as one XLA program for better compiler optimization.
Google announced new vLLM and SGLang inference engines built directly on PyTorch for TPU, developed with Inferact and RadixArk respectively. The engines run on a new TorchTPU layer that makes TPUs a native PyTorch device, replacing the previous JAX-based backend approach. Google reports a public benchmark result of 2.01 cents per million tokens in high-concurrency configurations via the SemiAnalysis InferenceX project, plus support for agentic workloads, long context, disaggregated serving and multimodal models.
Red Hat, IBM, and Google announced llm-d 1.0, a Kubernetes-native distributed inference serving system that moves beyond proof-of-concept performance tricks into full production readiness. The release adds multi-tenant quality of service, zero-downtime operations, and agent-aware serving so the same endpoint can handle both human users and autonomous agents efficiently. In one test, llm-d's prefix-aware router captured 97% of available cache reuse on an agentic coding workload, and IBM/Red Hat ran 3,000 concurrent software-engineering agents for 5 to 10 times less cost than using the same model as a hosted service.
A new study questions the common belief that whether an AI agent falls for a prompt injection is settled the moment it reads the malicious text. The researchers built a branching-rollout technique on AgentDojo benchmarks: they replayed an agent's run up to the point it encounters an injected instruction, then let the rest of the episode play out many different ways to measure how much of the final outcome was already locked in versus still open. For a Qwen3 agent running without extended reasoning, 91% of the outcome was already determined at that moment; switching on reasoning ('thinking') cut that figure to 62%, showing the agent's compliance decision often keeps forming well after it first sees the attack. Enabling reasoning also cut Qwen3's susceptibility to successful attacks, but made two other tested models more vulnerable, and benchmark scores for reasoning agents flipped on a rerun as often as 35% of the time.
AWS released a push-based NIXL connector for vLLM that changes how key-value (KV) cache data moves between prefill and decode instances in disaggregated LLM serving. Instead of the decode instance pulling KV blocks after prefill finishes, the decode instance registers its target memory in advance and the prefill instance writes the KV data directly once ready, moving coordination work off the critical path. Benchmarked on a 1-prefill/1-decode Qwen3-32B deployment using EC2 P5en instances with EFA, push mode delivered up to 3x better time-to-first-token (TTFT) and up to 30% better time-per-output-token (TPOT) than the default pull mode, especially under high request rates.
Microsoft and NVIDIA released a joint whitepaper, 'Risk Management for Open-Weight Models in Enterprise,' the first in a planned series offering enterprises a shared risk taxonomy and methodology for evaluating open-weight AI models. The paper argues that whether a model is open or closed-weight is not the key risk factor — what matters is the use case, deployment architecture, and surrounding safeguards. It organizes risks into five categories (Cybersecurity, Data Protection, Content, Human-AI Interaction, Bias) and maps evaluations to those risks, aligning with ISO standards and the NIST AI Risk Management Framework.
Google Research describes new work on grounding AI agents in physical lab reality, using a digital-twin framework called VERID to let agents safely plan and execute liquid-handling robot protocols. Combined with closed-loop molecular design cycles for aptamers and proteins, the agents iteratively learned from wet-lab results, improving hit rates over multiple rounds and discovering single-digit nanomolar protein binders for a cancer-related target.
Google Cloud published a reference architecture for building AI agents that coordinate long-running, GPU-heavy scientific tasks like protein folding and compound screening, which break standard single-LLM-agent setups due to timeouts and context overflow. The design separates a cognitive orchestrator (Gemini on Agent Runtime) from a GPU execution layer (NVIDIA NIMs like ESMFold and GenMol on GKE) and wraps both in an enterprise governance plane with agent identity, a gateway, a registry and Model Armor. It also lays out best practices such as passing bulky data by reference instead of inline, streaming progress updates, and gating irreversible actions behind human approval.
Reflection, a new US AI lab, has launched Beam, its first model, as an open-weight mixture-of-experts system with 501B total parameters (23B active per token) and a 1M-token context window, released under Apache 2.0. The model is built for coding and agentic use cases, ships with a custom Rust-based inference harness, and has day-0 support in transformers, vLLM and SGLang. Reflection reports strong results on coding, reasoning and tool-calling benchmarks across BF16, FP8 and NVFP4 checkpoints.
NVIDIA, working with IIT Madras's AI4Bharat lab, is releasing Nemotron-pretraining-Indic-v1, a roughly 2.8 trillion token synthetic pretraining corpus covering 22 Indian languages, along with the open GPU-accelerated pipeline used to generate it. The corpus includes native-script data for all 22 languages and romanized text for twelve, built by translating curated English source material (web text, math, code, Q&A) using NeMo Data Designer and quality-checked with an MQM-inspired rubric. NVIDIA already used a 10-billion-token Indic subset for continued pretraining of Nemotron 3.5 Super VL, which showed measurable gains on Indic benchmarks.
IBM has released Granite Switch 4.1 in 8B and 30B sizes, a modular system that runs chat, decision-making and speech transcription from one hosted endpoint instead of separate models. The update adds a general-purpose Decision adapter, a specialized Guardian adapter for safety-related classifications, and an embedded ASR model, Granite-Speech-5.0-Turbo-CTC, which tops the OpenASR Leaderboard. Despite bundling three capabilities, the combined model uses roughly the same VRAM as the base model alone.
Meta published Muse Spark Open, an open-weight mixture-of-experts model with about 816 billion total parameters and 42 billion active per token. It accepts text, images, audio and video as input and generates text, with a context window of up to 1 million tokens. Meta positions it as a coding-optimized, agentic model built on top of the smaller Muse Glimmer architecture, calling it the most capable US open-weights model released so far.
Apple's MLX team published benchmarks showing large open-weight models — including a near-trillion-parameter MoE model called Muse Spark Open — running on Mac Studio hardware, both on a single machine and distributed across up to four Mac Studios using a new RDMA-over-Thunderbolt driver. The new M5 Ultra chip, with dedicated GPU Neural Accelerators and 1.2 TB/s memory bandwidth, delivers up to 4.6x faster time-to-first-token and 1.6x faster generation speed versus the previous M3 Ultra. For models too large to fit on one machine, MLX now supports pipeline and tensor parallelization across multiple Macs connected via Thunderbolt RDMA, letting a four-Mac-Studio cluster generate over 36 tokens per second on an 816B-parameter model.
Swiss AI released Apertus 1.5, a continued pretraining of its open Apertus 1.0 models (8B and 70B, Apache 2.0) with native image and experimental audio understanding, a 4x larger context window of 262,144 tokens, an optional thinking mode, and improved instruction-following and tool use. The models were trained on a total of 19T (8B) and 17T (70B) tokens, with full training data, recipes and intermediate checkpoints published for reproducibility. The team also published detailed, standardized evaluation cards and announced early API access from providers like Swisscom and Infomaniak, plus a roadmap for a ~600B-parameter MoE Apertus 2.0.
Researchers describe a system called FlowSpec that converts a creditor's legal call procedure into machine-readable instructions, tool restrictions, training signals and test cases for a voice agent. A 9-billion-parameter model was trained on calls covering nearly 300 randomly generated procedures, then tested on four procedures it had never encountered, including one entirely held-out family of scripts, to see whether compliance behavior transfers.