NEAR AI Joins Forces With Bittensor’s SayGm to Make Private AI Easier to Use

NEAR AI and SayGm (Subnet 28 on Bittensor) have joined forces to make AI more private and easier to access.

Announced on September 18, the collaboration brings NEAR AI’s private AI models to SayGm, allowing developers to access them through a single API key without exposing their conversations or sensitive information to either company.

NEAR AI Cloud's confidential inference is now live on @say_gm_'s confidential tier.

SayGm reaches dozens of models through one API key, and runs its own routing inside an Intel TDX enclave rather than on ordinary servers. pic.twitter.com/NZahxG71D3

— NEAR AI (@near_ai) September 18, 2026

The partnership also brings two decentralized AI ecosystems together, showing how networks can combine their strengths to build better AI services.

What are NEAR AI and SayGm?

NEAR AI is building AI infrastructure that puts users in control of their data. It runs AI models in secure computing environments designed to keep information hidden even from the companies operating the servers.

NEAR was co-founded by Illia Polosukhin, one of the researchers behind the Transformer architecture that helped make modern AI possible.

SayGm (Subnet 28) is Bittensor’s AI gateway. It gives developers access to different AI models through one API key, removing the need to connect to each provider separately.

Its privacy-focused service also uses secure hardware to protect the information passing through its system.

More about GM (SN28) below:

How gm (SN28) Protects AI Prompts With Hardware-Based Privacy

How the collaboration works

When someone sends a message to an AI model, that information usually passes through a service that routes it to the model. Both the routing service and the company running the model may be able to see it.

NEAR AI and SayGm are changing that.

SayGm protects the message as it travels to the AI model. NEAR AI protects it while the model processes it. Both use secure computing environments called Trusted Execution Environments (TEEs), which help keep the information hidden from both companies running the systems.

NEAR co-founder Illia Polosukhin stressed why this matters, saying that end-to-end encryption should be a requirement for any serious business. Without it, companies cannot be sure who handles their information along the way or whether someone could change their requests.

Developers don’t have to take NEAR’s and GM’s word for it. They can check hardware-generated proof to confirm the AI model is running in a secure environment.

Note that the protection applies to models in SayGm’s confidential tier. Models from providers such as OpenAI and Anthropic remain subject to those providers’ privacy terms.

Why this matters for decentralized AI

The partnership makes private AI easier to use. Developers can access NEAR AI’s protected models through SayGm’s familiar API without building their own secure computing systems.

It also shows how decentralized AI networks can work together.

NEAR AI provides the secure environment where AI models run, while SayGm gives developers a way to access them through Bittensor.

Mark Jeffrey, a partner at Stillcore Capital, welcomed the collaboration and argued that Bittensor can work alongside other decentralized AI projects instead of competing with them.

“We’re rowing in the same direction,” he said.

The collaboration brings that idea to life. Different networks can contribute different services, allowing developers to use them together without being tied to a single ecosystem.

For users, the benefit is greater access to AI services that protect their information. For developers, it means more models and stronger privacy through a familiar interface.

NEAR and SayGm are showing how decentralized AI can move forward through collaboration, making private AI more accessible while giving users greater control over their data.