NEAR AI Cloud becomes provider on OpenRouter, serving GLM 5.3 Flash with 1M token context
The integration brings confidential AI inference to one of the most popular model routing platforms, with privacy guarantees backed by hardware-level encryption.
NEAR AI Cloud is now live as a provider on OpenRouter, giving developers access to the GLM 5.3 Flash model through the provider slug near-ai. The model comes with a 1 million token context window, and NEAR’s implementation wraps it in something most AI providers don’t offer: cryptographically verified confidential inference.
What GLM 5.3 Flash actually is
The model being served isn’t homegrown. GLM 5.3 Flash was developed by Z.ai and publicly launched on August 26 after an anonymous preview period called “Ox Alpha.” It’s a multimodal mixture-of-experts architecture, which means it routes different parts of a query to specialized sub-networks rather than firing up the entire model for every request.
The numbers are substantial: 320 billion total parameters, with 18 billion active at any given time. It handles text, images, and video as inputs, producing text outputs.
The model’s open weights are released under the MIT license. On NEAR AI Cloud, the pricing sits at $0.15 per million input tokens and $0.50 per million output tokens.
The privacy angle isn’t just marketing
NEAR AI Cloud runs inference inside hardware-based Trusted Execution Environments, specifically Intel TDX and NVIDIA’s confidential computing solutions. Every request comes with a cryptographic attestation report, which is essentially a mathematical proof that the computation happened inside a secure enclave and that no one, including NEAR, could have accessed the data.
AI, tech, and the markets they move—in one daily briefing.
Daily. Free. Join 34,000+ readers across crypto, finance, and policy.
The platform maintains a zero data retention policy. Your prompts go in, your responses come out, and nothing gets stored or logged on NEAR’s side.
For OpenRouter users specifically, this means they can select the near-ai provider without needing to create a separate NEAR AI account.
Staking meets compute credits
The connection between NEAR’s blockchain layer and its AI infrastructure runs through the token’s staking mechanism. NEAR holders who stake their tokens can convert staking yields into compute credits for NEAR AI Cloud services.
You can use NEAR AI Cloud through OpenRouter without touching a single token. But if you’re already holding NEAR, your staking rewards now have a concrete use case beyond governance and speculative appreciation.