OpenAI Models Land on AWS Through Amazon Bedrock
Overview
OpenAI announced on April 28, 2026 that its GPT models and Codex coding agent are now available through Amazon Bedrock, AWS's managed AI platform. The deployment follows the prior day's announcement of the Microsoft-OpenAI partnership restructure ending Azure exclusivity, and represents OpenAI's first major multi-cloud rollout.
What's Available
- GPT Models: AWS customers can experiment with OpenAI's latest GPT models through Amazon Bedrock, with general availability expected in the coming weeks.
- Codex Agent: OpenAI's code-writing agent is also available via Amazon Bedrock Managed Agents, enabling developers to build sophisticated customized agents with persistent memory of previous interactions.
How It Happened
The AWS deployment is a direct result of Amazon's February 2026 commitment of up to $50 billion to OpenAI, made on condition that AWS would have rights to serve OpenAI's new agent-building tool, Frontier. This created a conflict with OpenAI's Azure exclusivity, leading to the partnership restructure that made multi-cloud deployment possible.
Significance
AWS holds roughly 31% of the global cloud market, the largest share of any provider. Making OpenAI models available through Bedrock immediately puts them within reach of AWS's massive enterprise customer base, allowing companies to adopt OpenAI technology without migrating off existing AWS infrastructure. This significantly expands OpenAI's enterprise market reach beyond Azure's customer base.
Source: CNBC
Related Articles
Microsoft and OpenAI restructured their partnership on April 27, ending Azure exclusivity so OpenAI can serve products on any cloud. Microsoft stops paying revenue share to OpenAI, while OpenAI continues capped payments to Microsoft through 2030.
Microsoft and OpenAI restructured their partnership on April 27, ending Azure cloud exclusivity and removing the controversial AGI clause. OpenAI can now serve customers on any cloud platform and is free to pursue multi-cloud deals ahead of a potential 2026 IPO.
This matters because the next bottleneck in agent coding is human attention, not raw model speed. OpenAI says Symphony lifted landed pull requests by 500% on some teams after engineers hit a practical ceiling of roughly three to five concurrent Codex sessions.
Comments (0)
No comments yet. Be the first to comment!