OpenAI

About this provider

OpenAI was founded in 2015 in San Francisco. The company released GPT-3 in 2020 and launched ChatGPT in November 2022 — the moment AI chatbots broke into mainstream awareness. Since then the GPT family evolved from GPT-3.5 via GPT-4 and GPT-4o into the current GPT-5 series and the specialised o-series for reasoning.

OpenAI hosts all models on its own Microsoft Azure infrastructure in the United States. For customers with an EU data residency requirement we recommend the ZelixAI Privacy Cluster. For customers without that requirement, OpenAI offers the world's broadest model selection: from spot-cheap classification (GPT-4o mini) to pure reasoning (o3) to multimodal (GPT-4o) to the latest flagship (GPT-5.5).

Why ZelixAI uses this provider

• Broadest model offering: from $0.15/1M (mini) to premium reasoning, all in one account
• Native multimodal models — GPT-4o accepts text, images and audio
• Aggressive price drops since 2024 — often cheaper than Anthropic equivalents
• Customer data has not been used for training by default since March 2023

Models from this provider

The following models are currently available through this provider within ZelixAI:

How does ZelixAI handle your data?

Requests via the OpenAI route go through an encrypted API connection (TLS 1.3) to OpenAI's Azure infrastructure in the US. Since March 2023, OpenAI does not use customer data for model training by default (opt-out is the default for API customers). Only for functional use — chat history, knowledge-base searches and (soon) mailbox integrations — is your data temporarily stored in your own tenant database, under your control (view, export or delete via the portal). In addition, ZelixAI logs metadata (model, tokens, latency) for billing and debugging.