NEAR AI Cloud Is Now a Provider on OpenRouter

NEAR AI Cloud is now available as a provider on OpenRouter. If you already route through OpenRouter, our models are reachable with the key and the balance you have.

What is OpenRouter?

OpenRouter is a unified API for large language models: one endpoint, hundreds of models from dozens of providers, with routing, fallbacks and cost tracking across all of them.

It is a drop-in replacement for the OpenAI API, so switching models means changing a model identifier rather than an SDK. One key and one balance replace an account with every provider you want to reach. Inference carries no markup, since OpenRouter passes through each provider's own rate and takes its fee when you load credits instead.

What we serve there

NEAR AI Cloud appears under the provider slug near-ai. We serve one model today: GLM 5.3 Flash, with a 1M token context window.

GLM 5.3 Flash is a natively multimodal model from Z.ai, which OpenRouter lists as accepting text, image and video and returning text. Z.ai built it for efficient coding and long-horizon agent work.

The model runs on the same TEE hardware our direct API uses, but the gateway in front of this route is not attested. Read that as how we run our infrastructure, not as a guarantee about your request.

How to route to NEAR AI on OpenRouter

Providers serving the same model are not interchangeable. Precision, context window, price and data policy all vary between them, and OpenRouter picks for you unless you say otherwise.

One line puts us first:

{
  "model": "z-ai/glm-5.3-flash",
  "messages": [{ "role": "user", "content": "..." }],
  "provider": {
    "order": ["near-ai"]
  }
}

order tells OpenRouter who to try first. Fallbacks stay on by default, so you get us first and keep a route either way.

If you need us and nobody else, add "allow_fallbacks": false. The request then fails rather than landing with a provider you did not choose. Use it when only one provider will do, and turn it on deliberately.

If you would rather set it in the interface, there is a short walkthrough of selecting NEAR AI as your provider on OpenRouter.

For confidential inference, come to NEAR AI directly

Your request travels through OpenRouter's infrastructure and then through a gateway outside our confidential environment, so this route is not confidential inference. It is a fine deal when what you want is one key across hundreds of models.

When you need confidentiality, skip the router. Call the NEAR AI Cloud API directly and your prompt is processed inside Intel TDX virtual machines paired with NVIDIA GPUs in confidential-computing mode. You can check the attestation report against Intel and NVIDIA rather than against us, and the report is free and needs no account.

What a team gets on the direct API:

• Frontier and confidential open-weight models
• One organization, workspaces for each team
• One API key per member, each with its own spend limit and usage reporting
• Pay per token. No subscriptions, no seat licenses
• Free attestation reports, no account needed

Start at cloud.near.ai, or read the verification docs first.

Frequently asked questions

How do I use NEAR AI on OpenRouter?

Select NEAR AI as the provider. Our slug is near-ai. Set "provider": {"order": ["near-ai"]} in the request body, or choose us in the OpenRouter interface. Leave fallbacks on and you keep a route either way. You do not need a separate NEAR AI account.

Which NEAR AI models are on OpenRouter?

One today: GLM 5.3 Flash. We run more on our own API, including embedding and reranker models, image generation and speech to text. Browse them here.

Is my prompt confidential if I route through OpenRouter?

No. Requests arriving through a router pass through infrastructure outside our confidential environment. For confidential inference, call the NEAR AI Cloud API directly. Get a key at near.ai.

How do I get confidential inference from NEAR AI?

Use the NEAR AI Cloud API with a key from near.ai. Prompts and workloads are processed inside a trusted execution environment, and the attestation report is free and available without an account.

Does OpenRouter add a markup on inference?

No. OpenRouter passes through each provider's own pricing and charges its fee when you load credits instead.