- Deployment
- hosted
- Hq Country
- United States
- Hq City
- Palo Alto
- Ownership
- private
- Founded
- 2022
- Training Default
- never
- Residency Options
- not published
- Gov Access Exposure
- us-cloud-act
- Eu Ai Act Role
- deployer
- Certs
- ISO 27001, SOC 2 Type II, HIPAA
- DPA available
- no
- Zero Retention Available
- Yes
- Retention Default
- "Input data is not stored to disk during inference — it exists only in memory while the request is being processed. Output data is not stored — it is sent to you and then deleted from memory. We do not train on your data. We do not share your data with third parties."
- Retention Exceptions
- Three carve-outs, all disclosed on the same page: image-generation outputs are stored briefly for the model demo page; bulk inference APIs may store data on disk in encrypted form until results are returned; and if you select a Google or Anthropic model, that vendor stores the output under its own terms.
- Subprocessors
- Stripe, Amazon Web Services, Google Cloud Platform
- Subprocessor Count
- 3
- Transfer Mechanism
- none-stated
- Eu Procurement Ready
- fail
- Eu Procurement Reason
- The clearest inference data-handling page of any vendor in this batch — in-memory only, no training, with the exceptions written down rather than buried. Everything around it is weaker: the subprocessor list has three entries and was last updated September 2024, no Article 46 transfer mechanism is named, no DPA is published, no EU residency exists, and the trust center self-declares "SOC 2 Compliant" and "ISO 27001 Compliant" with GDPR marked "In progress" — wording that is not the same as holding a certificate.
- Training Claim Basis
- stated
- Training Confidence
- high
- Residency Claim Basis
- inferred
- Residency Confidence
- medium
- Retention Claim Basis
- stated
- Retention Confidence
- high
- Subprocessors Claim Basis
- stated
- Subprocessors Confidence
- high
- Residency Class
- non-eu-only
- Portability / Exit Path
- serves-open-models
- Subprocessor Jurisdiction
- non-eu
- Underlying model providers
- Meta (Llama), DeepSeek, Qwen, Mistral AI, Google, Anthropic
