Enterprise security teams drown in alerts but struggle to fix known risks. AI agents are changing how organizations remediate vulnerabilities at scale.
Marshall has announced a new and smaller party speaker. The classic guitar amplifier brand’s Bromley 450 Bluetooth speaker includes instrument and microphone inputs.
Valorant gets a new agent as part of its latest set of patch notes, but significant nerfs to others will no doubt make some players unhappy.
Social media influencers are spreading brand-related misinformation in a much more toxic way than ordinary users, research has found.
High-end audio brand Astell&Kern is launching the Activo Scoop, a new pair of wired earphones for audiophiles on a budget, just as wired IEMs are undergoing a resurgence.
Two of Apple’s most beloved iPhone designs have been made obsolete in the last few hours. Check yours now (and the one in that special drawer).
Every few years, a piece of open-source software arrives that rewires how the industry thinks about computing. Linux did it for servers. Docker did it for deployment. OpenClaw — the autonomous AI agent platform that went from niche curiosity to the fastest-growing open-source project in history in a matter of weeks — may be doing it for software itself.
Nvidia CEO and co-founder Jensen Huang made his position plain at GTC 2026 this week: “OpenClaw is the operating system for personal AI. This is the moment the industry has been waiting for — the beginning of a new renaissance in software.” And Nvidia wants to be the company that makes it enterprise-ready.
At its annual large GTC 2026 conference in San Jose this week, Nvidia unveiled NemoClaw, a software stack that integrates directly with OpenClaw and installs in a single command. Along with it came Nvidia OpenShell, an open-source security runtime designed to give autonomous AI agents — or “claws”, as the industry is increasingly calling them — the guardrails they need to operate inside real enterprise environments. Alongside both, the company announced an expanded Nvidia Agent Toolkit, a full-stack platform for building and running production-grade agentic workflows.
The message from Jensen Huang was unambiguous. “Claude Code and OpenClaw have sparked the agent inflection point — extending AI beyond generation and reasoning into action,” the Nvidia CEO said ahead of the conference. “Employees will be supercharged by teams of frontier, specialized and custom-built agents they deploy and manage.” Watch my video overview of it below and read on for more:
The terminology shift happening inside enterprise AI circles is subtle but significant. Internally, teams building with OpenClaw and similar platforms have taken to calling individual autonomous agents claws — a nod to the platform name, but also a useful shorthand for a new class of software that differs fundamentally from the chatbots and copilots of the last two years.
As Kari Briski, Nvidia’s VP of generative AI software, put it during a Sunday briefing: “Claws are autonomous agents that can plan, act, and execute tasks on their own — they’ve gone from just thinking and executing on tasks to achieving entire missions.”
That framing matters for IT decision-makers. Claws are not just assistants. They are persistent, tool-using programs that can write code, browse the web, manipulate files, call APIs, and chain actions together over hours or days without human input. The productivity upside is substantial. So is the attack surface. Which is precisely the problem Nvidia is positioning NemoClaw to solve.
The enterprise demand is not hypothetical. Harrison Chase, founder of LangChain — whose open-source agent frameworks have been downloaded more than a billion times — put it bluntly in a recent episode of VentureBeat’s Beyond the Pilot podcast: “I guarantee that every enterprise developer out there wants to put a safe version of OpenClaw onto onto their computer or expose it to their users.” The bottleneck, he made clear, has never been interest. It has been the absence of a credible security and governance layer underneath it. NemoClaw is Nvidia’s answer to that gap — and notably, LangChain is one of the launch partners for the Agent Toolkit and OpenShell integration.
NemoClaw is not a competitor to OpenClaw (or the now many alternatives). It is best understood as an enterprise wrapper around it — a distribution that ships with the components a security-conscious organization actually needs before letting an autonomous agent near production systems.
The stack has two core components. The first is Nvidia Nemotron, Nvidia’s family of open models, which can run locally on dedicated hardware rather than routing queries through external APIs. Nemotron-3-Super, scored the highest out of all open models on PinchBench, a benchmark that tests the types of tasks and tools calls needed by OpenClaw.
The second is OpenShell, the new open-source security runtime that runs each claw inside an isolated sandbox — effectively a Docker container with configurable policy controls written in YAML. Administrators can define precisely which files an agent can access, which network connections it can make, and which cloud services it can call. Everything outside those bounds is blocked.
Nvidia describes OpenShell as providing the missing infrastructure layer beneath claws — giving them the access they need to be productive while enforcing policy-based security, network, and privacy guardrails.
For organizations that have been watching OpenClaw’s rise with a mixture of excitement and dread, this is a meaningful development. OpenClaw’s early iterations were, by general consensus, a security liability — powerful and fast-moving, but essentially unconstrained. NemoClaw is the first attempt by a major hardware vendor to make that power manageable at enterprise scale.
One aspect of NemoClaw that deserves more attention than it has received is the hardware strategy underneath it. Claws, by design, are always-on — they do not wait for a human to open a browser tab. They run continuously, monitoring inboxes, executing tasks, building tools, and completing multi-step workflows around the clock.
That requires dedicated compute that does not compete with the rest of the organization’s workloads. Nvidia has a clear interest in pointing enterprises toward its own hardware for this purpose.
NemoClaw is designed to run on Nvidia GeForce RTX PCs and laptops, RTX PRO workstations, and the company’s DGX Spark and DGX Station AI supercomputers. The hybrid architecture allows agents to use locally-running Nemotron models for sensitive workloads, with a privacy router directing queries to frontier cloud models when higher capability is needed — without exposing private data to those external endpoints.
It is an elegant solution to a real problem: many enterprises are not yet ready to send customer data, internal documents, or proprietary code to cloud AI providers, but they still need model capability that exceeds what runs locally. NemoClaw’s privacy router architecture threads that needle, at least in principle.
Before evaluating the platform, it helps to understand what a claw doing real work looks like in practice. Two partner integrations announced alongside NemoClaw offer the clearest window into where this is heading.
Box is perhaps the most illustrative case for organizations that manage large volumes of unstructured enterprise content.
Box is integrating Nvidia Agent Toolkit to enable claws that use the Box file system as their primary working environment, with pre-built skills for Invoice Extraction, Contract Lifecycle Management, RFP sourcing, and GTM workflows.
The architecture supports hierarchical agent management: a parent claw — such as a Client Onboarding Agent — can spin up specialized sub-agents to handle discrete tasks, all governed by the same OpenShell Policy Engine.
Critically, an agent’s access to files in Box follows the exact same permissions model that governs human employees — enforced through OpenShell’s gateway layer before any data is exchanged. Every action is logged and attributable; no shadow copies accumulate in agent memory. As Box puts it in their announcement blog, “organizations need to know which agent touched which file, when, and why — and they need the ability to revoke access instantly if something goes wrong.”
Cisco’s integration offers perhaps the most visceral illustration of what OpenShell guardrails enable in practice. The Cisco security team has published a scenario in which a zero-day vulnerability advisory drops on a Friday evening.
Rather than triggering a weekend-long manual scramble — pulling asset lists, pinging on-call engineers, mapping blast radius — a claw running inside OpenShell autonomously queries the configuration database, maps impacted devices against the network topology, generates a prioritized remediation plan, and produces an audit-grade trace of every decision it made.
Cisco AI Defense verifies every tool call against approved policy in real time. The entire response completes in roughly an hour, with a complete record that satisfies compliance requirements.
“We are not trusting the model to do the right thing,” the Cisco team noted in their technical writeup. “We are constraining it so that the right thing is the only thing it can do.”
Nvidia is not building this alone. The Agent Toolkit and OpenShell announcements came with a significant roster of enterprise partners — Box, Cisco, Atlassian, Salesforce, SAP, Adobe, CrowdStrike, Cohesity, IQVIA, ServiceNow, and more than a dozen others — whose integration depth signals how seriously the broader software industry is treating the agentic shift.
On the infrastructure side, OpenShell is available today on build.nvidia.com, supported by cloud inference providers including CoreWeave, Together AI, Fireworks, and DigitalOcean, and deployable on-premises on servers from Cisco, Dell, HPE, Lenovo, and Supermicro. Agents built within OpenShell can also continuously acquire new skills using coding agents including Claude Code, Codex, and Cursor — with every newly acquired capability subject to the same policy controls as the original deployment.
Separately, Nvidia announced the Nemotron Coalition — a collaborative initiative bringing together Mistral AI, Perplexity, Cursor, and LangChain to co-develop open frontier models. The coalition’s first project is a base model co-developed with Mistral that will underpin the upcoming Nemotron 4 family, aimed specifically at agentic use cases.
The NemoClaw announcement marks a turning point in how enterprise AI is likely to be discussed in boardrooms and procurement meetings over the next twelve months. The question is no longer whether organizations will deploy autonomous agents. The industry has clearly moved past that debate. The question is now how — with what controls, on what hardware, using which models, and with what audit trail.
Nvidia’s answer is a vertically integrated stack that spans silicon, runtime, model, and security policy. For IT leaders evaluating their agentic roadmap, NemoClaw represents a significant attempt to provide all four layers from a single vendor, with meaningful third-party security integrations already in place.
The risks are not trivial. OpenShell’s YAML-based policy model will require operational maturity that most organizations are still building. Claws that can self-evolve and acquire new skills — as Nvidia’s architecture explicitly enables — raise governance questions that no sandbox can fully resolve. And the concentration of agentic infrastructure in a single vendor’s stack carries familiar platform risks.
That said the direction is clear. Claws are coming to the enterprise. Nvidia just made its bet on being the platform they run on — and the guardrails that keep them in bounds.
Leaked production figures show Apple is betting big on the iPhone Fold, dwarfing Samsung’s 2026 targets. But at $2000, it’s a big gamble.
Apple has let slip when the next iPhone software will arrive in the fine print of its new AirPods Max 2 reveal.
Chinese AI startup Z.ai, known for its powerful, open source GLM family of large language models (LLMs), has introduced GLM-5-Turbo, a new, proprietary variant of its open source GLM-5 model aimed at agent-driven workflows, with the company positioning it as a faster model tuned for OpenClaw-style tasks such as tool use, long-chain execution and persistent automation.
It’s available now through Z.ai’s application programming interface (API) on third-party provider OpenRouter with roughly a 202.8K-token context window, 131.1K max output, and listed pricing of $0.96 per million input tokens and $3.20 per million output tokens. That makes it about $0.04 cheaper per total input and output cost (at 1 million tokens) than its predecessor, according to our calculations.
|
Model |
Input |
Output |
Total Cost |
Source |
|
Grok 4.1 Fast |
$0.20 |
$0.50 |
$0.70 |
|
|
Gemini 3 Flash |
$0.50 |
$3.00 |
$3.50 |
|
|
Kimi-K2.5 |
$0.60 |
$3.00 |
$3.60 |
|
|
GLM-5-Turbo |
$0.96 |
$3.20 |
$4.16 |
|
|
GLM-5 |
$1.00 |
$3.20 |
$4.20 |
|
|
Claude Haiku 4.5 |
$1.00 |
$5.00 |
$6.00 |
|
|
Qwen3-Max |
$1.20 |
$6.00 |
$7.20 |
|
|
Gemini 3 Pro |
$2.00 |
$12.00 |
$14.00 |
|
|
GPT-5.2 |
$1.75 |
$14.00 |
$15.75 |
|
|
GPT-5.4 |
$2.50 |
$15.00 |
$17.50 |
|
|
Claude Sonnet 4.5 |
$3.00 |
$15.00 |
$18.00 |
|
|
Claude Opus 4.6 |
$5.00 |
$25.00 |
$30.00 |
|
|
GPT-5.4 Pro |
$30.00 |
$180.00 |
$210.00 |
Second, Z.ai is also adding the model to its GLM Coding subscription product, which is its packaged coding assistant service. That service has three tiers: Lite at $27 per quarter, Pro at $81 per quarter, and Max at $216 per quarter.
Z.ai’s March 15 rollout note says Pro subscribers get GLM-5-Turbo in March, while Lite subscribers get the base GLM-5 in March and must wait until April for GLM-5-Turbo. The company is also taking early-access applications for enterprises via a Google Form, which suggests some users may get access ahead of that schedule depending on capacity.
z.ai describes GLM-5-Turbo as designed for “fast inference” and “deeply optimized for real-world agent workflows involving long execution chains,” with improvements in complex instruction decomposition, tool use, scheduled and persistent execution, and stability across extended tasks.
The release offers developers a new option for building OpenClaw-style autonomous AI agents, and serves as a signal about where model vendors think enterprise demand is heading: away from chat interfaces and toward systems that can reliably execute multi-step work.
That is now where much of the competition is moving, as well, especially among vendors trying to win developers and enterprise teams building internal assistants, workflow orchestrators and coding agents.
Z.ai’s materials frame GLM-5-Turbo as a model for production-like agent behavior rather than static prompt-response use.
The pitch centers on reliability in practical task flows: better command following, stronger tool invocation, improved handling of scheduled and persistent tasks, and faster execution across longer logical chains. That positioning puts the model squarely in the market for agents that do more than answer questions.
It is aimed at systems that can gather information, call tools, break down instructions and keep working through complex task sequences with less supervision.
Rather than a straightforward successor to GLM-5, GLM-5-Turbo appears to be a more execution-focused variant: tuned for speed, tool use and long-chain agent stability, while the base GLM-5 remains Z.ai’s broader open-source flagship.
GLM-5-Turbo appears especially competitive in OpenClaw scenarios such as information search and gathering, office and daily tasks, data analysis, development and operations, and automation. Those are company-supplied materials, not independent validation, but they make the intended product positioning clear.
Founded in 2019 as a Tsinghua University spinoff in Beijing, Z.ai — formerly Zhipu AI — is now one of China’s best-known foundation model companies. The company remains headquartered in Beijing and is led by CEO Zhang Peng
Z.ai listed on the Hong Kong Stock Exchange on January 8, 2026, with shares priced at HK$116.20 and opening at HK$120, for a stated market capitalization of HK$52.83 billion, making it China’s largest independent large language model developer.
As of September 30, 2025 its models had reportedly been used by more than 12,000 enterprise customers, more than 80 million end-user devices and more than 45 million developers worldwide.
Z.ai’s last major release, GLM-5, which debuted in February 2026, gives useful context for what the company is now trying to do with GLM-5-Turbo.
GLM-5 is an open-source flagship model carrying an MIT license, posting a record-low hallucination score on the AA-Omniscience Index, and debuted a native “Agent Mode” that could turn prompts or source materials into ready-to-use .docx, .pdf and .xlsx files.
That earlier release was also framed as a major technical step up for the company. GLM-5 scaled to 744 billion parameters with 40 billion active per token in a mixture-of-experts architecture, used 28.5 trillion pretraining tokens, and relied on a new asynchronous reinforcement-learning infrastructure called “slime” to reduce training bottlenecks and support more complex agentic behavior.
In that light, GLM-5-Turbo looks less like a replacement for GLM-5 than a narrower commercial offshoot: a variant that keeps the long-context, agentic orientation of the flagship line but emphasizes speed, stability and execution in real-world agent chains.
On the technical side, Z.ai has been packaging the GLM-5 family with the kinds of capabilities developers now expect from serious agent-facing models, including long context handling, tools, reasoning support and structured integrations.
OpenRouter’s GLM-5-Turbo page lists support for tools, tool choice and response formatting, while also surfacing live performance data including average throughput and latency.
OpenRouter’s provider telemetry adds a useful deployment-level comparison between GLM-5 and GLM-5-Turbo, though the data is not perfectly apples-to-apples because GLM-5 appears across several providers while GLM-5-Turbo is shown only through Z.ai.
On throughput, GLM-5-Turbo averages 48 tokens per second on OpenRouter, which puts it below the fastest GLM-5 endpoints shown in the screenshots, including Fireworks at 70 tok/s and Friendli at 58 tok/s, but above Together’s 40 tok/s.
On raw first-token latency, GLM-5-Turbo is slower in the available data, posting 2.92 seconds versus 0.41 seconds for Friendli’s GLM-5 endpoint, 1.00 second for Parasail and 1.08 seconds for DeepInfra.
But the picture improves on end-to-end completion time: GLM-5-Turbo is shown at 8.16 seconds, faster than the GLM-5 endpoints, which range from 9.34 seconds on Fireworks to 11.23 seconds on DeepInfra.
The most notable operational advantage is in tool reliability. GLM-5-Turbo shows a 0.67% tool call error rate, materially lower than the GLM-5 providers shown, where error rates range from 2.33% to 6.41%.
For enterprise teams, that suggests a model that may not win on initial responsiveness in its current OpenRouter routing, but could still be better suited to longer agent runs where completion stability and lower tool failure matter more than the fastest first token.
A ZClawBench radar chart released by z.ai shows GLM-5-Turbo as especially competitive in OpenClaw scenarios such as information search and gathering, office and daily tasks, data analysis, development and operations, and automation.
Those are company-supplied benchmark visuals, not independent validation, but they do help explain how Z.ai wants the two models understood: GLM-5 as the broader coding and open flagship, and Turbo as the more targeted agent-execution variant.
One notable caveat is licensing. Z.ai says GLM-5-Turbo is currently closed-source, but it also says the model’s capabilities and findings will be folded into its next open-source model release. That is an important distinction. The company is not clearly promising to open-source GLM-5-Turbo itself.
Instead, it is saying that lessons, techniques and improvements from this release will inform a future open model. That makes the launch more nuanced than a clean break from openness.
Z.ai’s earlier GLM strategy leaned heavily on open releases and open-weight distribution, which helped it build visibility among developers.
GLM-5-Turbo’s licensing posture also lands in a wider Chinese market context that makes the launch more notable than a simple product update.
In recent weeks, reporting around Alibaba’s Qwen unit has raised fresh questions about how China’s leading AI labs will balance open releases with commercial pressure.
Earlier this month, Qwen division head Lin Junyang stepped down, becoming the third senior Qwen executive to leave in 2026, even though Alibaba’s Qwen family remains one of the most prolific open-model efforts anywhere, with more than 400 open-source models released since 2023 and more than 1 billion downloads.
Reuters then reported on March 16 that Alibaba CEO Eddie Wu would take direct control of a newly formed AI-focused business group consolidating Qwen and other units, amid scrutiny over strategy, profitability and the brutal price competition surrounding open-model offerings in China.
Even without overstating those developments, they help frame the broader question hanging over the sector: whether the economics of frontier AI are starting to push even historically open-leaning Chinese labs toward a more segmented strategy.
That does not mean Chinese labs are abandoning open source. But the pattern is becoming harder to ignore: open models help drive adoption, developer goodwill and ecosystem reach, while certain high-value variants aimed at enterprise agents, coding workflows and other commercially attractive use cases may increasingly arrive first as proprietary products.
In that sense, GLM-5-Turbo fits a larger possible shift in China’s AI market, one that looks increasingly similar to the playbook used by OpenAI, Anthropic and Google in the U.S.: openness as distribution, proprietary systems as business.
Seen in that light, GLM-5-Turbo looks like more than a speed-focused product update. It may be another sign that parts of China’s AI sector are moving toward the same hybrid model already common in the U.S.: openness as distribution, proprietary systems as business.
That would not mark the end of open-source AI from Chinese labs, but it could mean their most strategically important agent-focused offerings appear first behind closed access, even if some of their underlying advances later make their way into open releases.
For developers evaluating agent platforms, that makes GLM-5-Turbo both a product launch and a useful signal. Z.ai is still speaking the language of open models. But with this release, it is also showing that some of its most commercially relevant work may arrive first as proprietary infrastructure for enterprise-grade agent systems.