NEAR AI Cloud has joined the OpenRouter platform, allowing developers to access the GLM 5.3 Flash model. The implementation features a 1 million token context window and provides cryptographically verified confidential inference.
The GLM 5.3 Flash model, developed by Z.ai, utilizes a multimodal mixture-of-experts architecture. It contains 320 billion total parameters, with 18 billion active per request, and supports text, image, and video inputs.
Pricing for the service is set at $0.15 per million input tokens and $0.50 per million output tokens. The platform operates under a zero data retention policy, ensuring no prompts or responses are stored.
Inference runs within hardware-based Trusted Execution Environments, specifically Intel TDX and NVIDIA confidential computing solutions. Each request includes a cryptographic attestation report to verify secure processing.
NEAR token holders can convert staking yields into compute credits for the service. However, OpenRouter users can access the provider without needing a separate NEAR AI account or holding tokens.
