OpenAI said on Monday it will officially launch its latest artificial intelligence models GPT-5.5 and GPT-5.4, and its coding agent Codex, on Amazon Web Services' generative AI service Amazon Bedrock.

The launch allows South Korean companies to use OpenAI's latest models in an AWS environment. Model usage will be integrated into existing AWS cloud commitments.

OpenAI said GPT-5.5 autonomously carries out multi-step tasks and is strong in writing and debugging code, data analysis and agent-based coding. It said GPT-5.4 has higher cost efficiency per token, making it suitable for large-scale production workloads. Both models support multiple languages, including Korean.

OpenAI said more than 5 million people use Codex each week. It automates code writing, refactoring, debugging and testing.

Amazon Bedrock Codex can be used by linking with the Codex app, CLI and major integrated development environments such as Visual Studio Code, JetBrains and Xcode. Inference is provided through Amazon Bedrock. It can be adopted with per-token billing without developer-specific commitments.

For security, the company said AWS Nitro-based Zero Operator Access is applied, meaning even operations staff cannot access customer prompts or responses.

Kyeonghoon Kim (김경훈), head of OpenAI Korea, said, "It will be possible to quickly expand the scope of using OpenAI models based on existing systems and workflows."

Kiho Ham (함기호), head of AWS Korea, said, "South Korean customers can now immediately adopt OpenAI's latest models through Bedrock in their existing AWS environment."