Your data is already encrypted at rest and in transit. But what about in use? Introducing Confidential Computing in Model Vault: where nothing and no one can access your workloads (yes, not even us).

Confidential Computing keeps your data from being tracked or used in three ways: - End-to-end encrypted inference - Hardware-enforced isolation that extends to the GPU with @nvidia Confidential Computing - The ability to request a signed attestation token that you can independently verify against the hardware vendor's public keys, available at any time Your data under your control.

Confidential Computing is now available in Model Vault, and we are opening early access to a limited group of beta customers. Beta places are limited. Learn more here:

Fantastic - great release @cohere team!

💚

well done on launch guys. when’s the new model coming 😋

Amazing. We need more CC AI solutions. Our app only talks to CC models, would love to get you guys in there and support what you do.

love the video!

encrypted at rest is table stakes. does in-use cover the KV cache and tool outputs, or only the weights on disk? that's what I'd check before a customer prompt goes in

@NVIDIAAI This is a good product

Finally TEEs in prod!

How do you track misuse

静态加密解决不了运行时访问，这个区分很关键。模型能看到什么、能不能把数据带出去，最好都做成可追踪的权限和审计事件。

Confidential inference becomes operationally credible when attestation is part of the request path. Bind the signed measurement to model version, GPU policy, code digest, and retention rules so a customer can verify the enclave they actually used.

Great to see this after reading the spyware TOS of another model.

Is this just CC off nvidia hardware with a model inside the TEE ?

I don't use cohere but I like cohere

This is the kind of trust boundary that matters in practice: teams need to know not only that data is protected in transit and at rest, but who can access it during use. Making that legible turns privacy from a promise into an operating property.