Amazon Web Services (AWS) has announced the general availability of OpenAI’s premier frontier models, GPT-5.5 and GPT-5.4, alongside the Codex coding agent on its Amazon Bedrock platform. This rollout marks the official transition from a limited preview phase that began on April 28, following the end of OpenAI’s exclusive cloud arrangement with Microsoft. Enterprise customers can now access these tools natively within the Bedrock model catalog, utilizing the same unified APIs already used to deploy models from providers like Anthropic and Meta.
The language models are optimized for complex, autonomous, and multi-step tasks. GPT-5.5 is tailored for highly demanding cognitive workloads, while GPT-5.4 is positioned as the ideal option for price-to-performance efficiency. AWS is offering both models through the Bedrock Responses API at the exact same per-token rates as direct OpenAI pricing, with no additional platform fees. Crucially for enterprise budgets, all model consumption automatically applies toward a customer’s pre-existing AWS cloud spending commitments.
In tandem with the language models, the Codex software engineering agent has achieved general availability under a flexible, pay-per-token structure. All inference for Codex is routed directly through AWS infrastructure, supporting development workflows across major integrated development environments (IDEs) including Visual Studio Code, JetBrains, and Xcode via the Codex App and CLI.
