OpenAI on the 2nd announced the official launch of its artificial intelligence (AI) models GPT-5.5 and GPT-5.4, and the coding agent Codex, on Amazon Web Services (AWS)' fully managed Generative AI service Amazon Bedrock.
On Amazon Bedrock, OpenAI models run on Bedrock's inference engine and on AWS security and governance frameworks. With automatic capacity management, they maintain predictable response speeds under load, and, according to the company, request states are continuously preserved so that work can proceed without interruption even during hardware failures or node restarts.
In Bedrock's security environment, there is no remote login or operator access path at all, preventing unauthorized access to customers' prompts or responses. Customer data is not used for model training and is not shared with model providers.
With this launch, corporations can more easily use OpenAI's latest models within their existing AWS environments. Model usage is integrated into AWS Cloud commitments, so no separate procurement or billing procedures are required.
OpenAI Korea head Kim Kyung-hoon said, "Through Amazon Bedrock, corporations can use OpenAI models in the AWS environment with security, governance, and scalability," adding, "This will quickly broaden the scope of OpenAI model use based on existing systems and workflows."