AWS Bedrock adds OpenAI models, Codex, and managed agents service following revised Microsoft agreement
AWS has added OpenAI's latest models, Codex, and a new managed agents service to its Bedrock platform, one day after OpenAI revised its agreement with Microsoft. The integration follows OpenAI's up-to-$50 billion deal with Amazon.
AWS Bedrock adds OpenAI models, Codex, and managed agents service following revised Microsoft agreement
AWS's Bedrock service now includes OpenAI's latest models, its Codex code-writing service, and a new product called Bedrock Managed Agents, Amazon announced Tuesday. The integration came one day after OpenAI and Microsoft revised their agreement to end Microsoft's exclusive rights to OpenAI products.
Bedrock is Amazon's AI application building and model-choosing service. The new offerings became available following OpenAI's up-to-$50 billion deal with Amazon, which had been blocked by Microsoft's previous exclusivity arrangement.
What's included
The AWS integration includes:
- OpenAI's latest models (specific versions not disclosed)
- Codex code-writing service
- Bedrock Managed Agents, designed specifically for OpenAI's reasoning models
Bedrock Managed Agents offers features including agent steering and security controls, according to Amazon's blog post.
Shifting cloud partnerships
The announcement marks a significant shift in AI cloud partnerships. Amazon CEO Andy Jassy called the revised Microsoft-OpenAI agreement "a very interesting announcement" in a tweet Monday.
According to TechCrunch, the Microsoft-OpenAI relationship has been deteriorating, with both companies finding alternative partnerships. OpenAI has turned to AWS and Oracle, while Microsoft has partnered with Anthropic and is developing a new agent offering powered by Claude.
Pricing for OpenAI models on AWS Bedrock was not disclosed in the announcement.
What this means
This integration gives AWS customers direct access to OpenAI's models through Bedrock's unified interface, eliminating the need to work through Azure. It represents Amazon's aggressive push into the AI infrastructure market and validates its $50 billion investment in OpenAI. The speed of the integration—just one day after the Microsoft agreement revision—suggests Amazon had been preparing for this moment. The managed agents service specifically targeting reasoning models indicates AWS is positioning itself for the emerging agentic AI market, directly competing with Microsoft's Claude-based agent offerings.
Related Articles
Amazon Nova 2 Sonic Unifies Speech Recognition, Reasoning, and TTS in Single Streaming Model
Amazon Web Services released technical guidance for migrating text agents to voice assistants using Amazon Nova 2 Sonic, a native speech-to-speech model that combines automatic speech recognition, reasoning, tool calling, and text-to-speech in a single bidirectional streaming interface. The model supports asynchronous tool calling and built-in voice activity detection for handling interruptions.
GitHub Copilot switches to metered token billing June 1 as flat-rate model proves unsustainable
Microsoft's GitHub is ending flat-rate billing for Copilot on June 1, 2026, switching to usage-based metered tokens after acknowledging the request-based model is no longer sustainable. Copilot Pro subscribers ($10/month) will receive 1,000 GitHub AI Credits monthly, with each credit worth $0.01.
GitHub Copilot switches to token-based pricing June 1, ending unlimited usage model
GitHub Copilot transitions to token-based pricing effective June 1, 2026, replacing its premium request unit system. Base subscription prices remain unchanged at $10/month for Pro and $39/month for Pro+, but users now receive equivalent monthly AI Credits that deplete with usage—and service stops when credits run out.
Anthropic tests AI agent marketplace with 186 deals totaling $4,000 among employees
Anthropic conducted an internal experiment called Project Deal where AI agents represented 69 employees as buyers and sellers in a classified marketplace. The agents completed 186 real transactions totaling over $4,000, revealing that more advanced models achieved better outcomes but users couldn't detect the performance disparity.
Comments
Loading...