NVIDIA's BioNeMo team, working with the University of Washington's Institute for Protein Design, has introduced RFD4-Proteina, a model that generates both a protein's sequence and its full atomic structure in a single pass instead of handing off to a separate inverse-folding tool. By working with a residue-level representation and a streamlined diffusion process, the model needs only about 60 denoising steps instead of roughly 400, which translates into markedly quicker runs and a much smaller memory footprint than its predecessor RFD3 when generating a 200-residue protein. During generation, the model can also run a reward-guided search that scores and branches partial designs using GPU-accelerated evaluation tools such as OpenFold3 and TMol, steering computation toward the most promising candidates. It ships both as an open-source model on GitHub and as a NIM microservice reachable over HTTP, with toolkit skills available to let automated agents chain generation, scoring and evaluation steps together.
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%.
Google Cloud and NVIDIA announced general availability of A5X instances built on NVIDIA Vera Rubin NVL72 GPUs, ConnectX-9 SuperNICs and Google's Virgo network, targeting large-scale agentic inference and reinforcement learning. Google Cloud plans to scale Vera Rubin infrastructure to up to 960,000 Rubin GPUs across multiple sites, with general availability next month and planned expansion to Europe. London-based Ineffable Intelligence, founded by AlphaGo creator David Silver, is the first customer, using the platform for continuous reinforcement-learning 'superlearner' systems. WPP also demonstrated a creative agent architecture on A5X using NVIDIA Nemotron models and OpenShell, running workloads on a single instance that previously needed a multi-node cluster.
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.
NVIDIA has published OpenShell, an open-source runtime that treats agent workloads as untrusted and grants access to files, tools and systems only when needed. It pairs with the new NVIDIA Vera CPU and, optionally, BlueField-4 DPUs with NVIDIA Sentry for hardware-enforced monitoring. NVIDIA says this combination forms a full reference architecture for running autonomous agents safely at data-center scale.
NVIDIA has detailed Sentry, a new security framework that sits outside an AI agent's reach, enforcing policy and watching behavior from a dedicated hardware trust domain on BlueField-4 DPUs. The idea is simple: even if an agent is compromised or goes off-script while using tools and credentials, Sentry keeps watching and can cut it off because it doesn't depend on the host OS the agent runs on. Two components do the work — DOCA Tower, a proxy that inspects and authorizes every agent network request, and DOCA Argus, which reads live memory to understand what's actually running. When something looks wrong, the system can quarantine the agent within milliseconds, without needing the agent or the host to cooperate.
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.
Amazon partnered with the independent nonprofit AVERI to audit the quality of its own internal safety-evaluation datasets for its AI models, not just the models' outputs. AVERI reviewed a random, anonymized sample of safety-test prompts covering areas like harmful content, cybersecurity, privacy and copyright. The review confirmed strong topical coverage, diverse testing methods and low contamination risk, but flagged gaps in multilingual coverage and multistep adversarial testing. Amazon responded with four concrete changes to its evaluation process.
AWS introduced Agent Registry, a managed catalog service for Model Context Protocol (MCP) servers, agents, and agent skills. Instead of hardcoding every tool into an orchestrator, agents can now search an approved, versioned catalog at runtime and build a client from the returned descriptor. A reference sample demonstrates a Discover → Instantiate → Execute loop using Amazon Bedrock AgentCore Gateway and AgentCore runtime, with Anthropic Claude Sonnet as the foundation model.
Eon, a data protection ISV running on AWS, replaced per-tool credentials and local MCP servers with a single Amazon Bedrock AgentCore MCP Gateway. The gateway uses Cedar-based policies to enforce least-privilege access for both human engineers and an autonomous SRE agent running Claude on Amazon Bedrock, routing calls to MCP servers hosted on AgentCore Runtimes across multiple AWS accounts. Developer onboarding time dropped from about two days to about two minutes after rolling the system out to 100+ developers and roughly 10,000 sessions.
AWS and Keyfactor published a reference design that fixes a known weak point in Model Context Protocol (MCP) deployments: MCP has no built-in way to authenticate which agent is actually making a request. The architecture adds three trust layers — user-to-system, agent-to-agent, and agent-to-tool-server — built on signed request envelopes, delegation assertions signed by the user, RFC 8693 audience-restricted tokens, and live OCSP revocation checks. A working example runs on Amazon Bedrock and the AWS Strands Agents SDK with Keyfactor EJBCA issuing and validating the X.509 certificates, and the full code is on GitHub.
Mistral AI released Voxtral 3, a set of speech recognition and speech synthesis models paired with an SDK that lets a single real-time pipeline listen, reason, call tools and speak back. The recognition models (Small, Mini and Realtime Transcribe 3) cover 50+ languages with speaker diarization, keyword-biasing for technical vocabulary, and transcription latency under 500ms, which Mistral claims beats Whisper, Scribe v2 and Gemini 3.5 Transcribe on its benchmarks. The speech synthesis model, Mini TTS 3, handles 35+ languages and 50+ dialects with emotion, pacing and emphasis that adapt to context, priced at $13 per million characters, and a separate Pro Voice feature can reproduce a person's voice and emotional range from just 30 minutes of recorded audio. All components can run through Mistral's EU-hosted API or be deployed on a customer's own servers.
IBM Research's ALTK-Evolve extracts short behavioral rules from an agent's own training-task attempts and appends them to the system prompt of a basic ReAct agent running on GPT-4.1. On the τ²-bench customer-service benchmark (retail and airline domains), this learned-memory approach raised accuracy and reliability without updating model weights or using human annotation. The full 'lossless' guideline set pushed a GPT-4.1 agent to a retail Pass^4 score of 67.5%, topping the October 12, 2026 leaderboard snapshot, including beating a model fine-tuned specifically on synthetic retail data. The technique also improved a frontier GPT-5.5 agent on airline tasks, taking it to the top of that leaderboard as well.
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.
Google Research announced the final milestone of its Coral NPU open hardware standard — a 32-bit RISC-V architecture with a new Matrix Execution Unit for up to 1 TOPS at roughly 10mW — plus a v1.0 open-source IREE/MLIR compiler that takes JAX, PyTorch and TensorFlow models directly to the hardware. Synaptics is launching the first commercial silicon built on this standard, including the Astra SR200 microcontroller series (400 GOPS) and new Coralboard SR and SL developer kits, with the SL now generally available via DigiKey. Google is also partnering with VeriSilicon and GlobalFoundries to bring the open RTL into production chips.
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 has rolled out structured outputs across every Anthropic Claude model hosted on Amazon Bedrock, reaching both the standard runtime endpoint and the newer Mantle endpoint built for agentic workloads. Instead of hoping the model returns valid JSON, developers can now lock responses to an exact schema through constrained decoding, which blocks the model from generating tokens that would break the format. A companion feature, strict tool use, applies the same guarantee to function-calling arguments so agent workflows don't fail on malformed inputs.
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 announced an expansion of its Project Perception security framework to address risks from autonomous AI agents, framing them as a new category of 'insider' that needs distinct identity, permission, and risk controls. The update introduces Dynamic Permissions (least-privilege, task-scoped, time-bound access), Dynamic Agent Risk (continuous risk scoring based on behavior drift and context), and Dynamic Conditional Access (graduated responses from narrowing access to full runtime isolation). Microsoft is also extending discovery and identity support (Microsoft Entra Agent ID) to 60 types of local agents, including Claude Code, OpenAI Codex, and GitHub Copilot, and introducing 'Guardian Agents' that monitor and respond to risky agent behavior at machine speed.
DeepSeek released DeepSeek-V4.1-Flash-1017, an upgraded version of its Flash model, accessible through the same API endpoint and pricing as before. The update brings notable gains across agentic, coding, and tool-use benchmarks while reducing task runtime and output length. DeepSeek also detailed its in-house search engine, which now powers both web/app search and a developer-facing Search & Fetch API.
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.