Back to collectionNEAR AI Adds DeepSeek V4.1 Flash with Incognito Routing
The live model uses NEAR AI’s shared-credential router but runs at a partner provider without model-level TEE attestation or end-to-end encryption.
NEAR AI listed DeepSeek V4.1 Flash as a live reasoning and coding model with image understanding on October 22. The catalog exposes it through the NEAR AI Cloud API and says it is served through OpenRouter using zero-data-retention endpoints. NEAR AI’s shared provider credentials keep the customer’s NEAR AI account credentials from being forwarded to the model provider.
The privacy boundary is narrower than NEAR AI’s fully private models. The catalog classifies this route as Incognito and states that the partner provider processes prompts under its own terms. It provides HTTPS and gateway verification where available, but no model-level trusted-execution-environment attestation or client-to-model end-to-end encryption. Availability is therefore established; confidential-hardware execution is not.