OpenAI (ChatGPT)
External APIs
OpenAI offers the ChatGPT family and GPT models as SaaS, at the forefront for reasoning, code and multimodal capabilities. On-demand access, token-based billing, data subject to the vendor's policy. The platform uses it for non-sensitive innovation, as overflow from the local core.
Key points: ChatGPT and GPT · Reasoning and code · Token-based · Non-sensitive.
Models served: GPT-5.6 (Sol, Terra, Luna), GPT-5.5, GPT-5.4 and GPT-5.4 mini; GPT-4.1 and the o-series remain on the price list as earlier models; image (GPT-Image-2) and audio (OpenAI pricing page as of 2 September 2026)
Location: United States, subject to the Cloud Act.
Where the data goes
By default, processing happens in the United States, under the Cloud Act. OpenAI offers data residency endpoints, including a Europe zone (EEA plus Switzerland), billed at a 10 % uplift on list pricing for models released on or after 5 March 2026 (OpenAI documentation as of 2 September 2026). A European company that wants guaranteed processing in Europe more often goes through Azure OpenAI, which deploys the same models in EU regions.
Retention and training
API inputs and outputs are retained for up to thirty days in abuse monitoring logs, then deleted, and they are not used to train the models by default (OpenAI documentation as of 2 September 2026). Zero data retention exists but is negotiated: it is open only to eligible customers and covers only certain endpoints.
When to use it from a sovereign platform
For non-sensitive innovation and overflow capacity, when the local foundation covers daily work and only peaks justify an API. Not for health data, non-pseudonymised personal data or industrial secrets, unless zero data retention and regional processing are written into the contract.