AWS launches OpenAI’s first open-weight models in 6 years

AWS has launched day-of-launch availability of two new open-weight models from OpenAI on Amazon Bedrock and Amazon SageMaker, breaking Microsoft’s traditional exclusivity with OpenAI. This marks the first time OpenAI has released open-weight models since GPT-2 in 2019, allowing AWS customers to fine-tune the models for specific use cases without directly interacting with OpenAI.

What you should know: The two new models—gpt-oss-120b and gpt-oss-20b—represent OpenAI’s first open-weight releases in six years.

• Open-weight models have visible parameters that allow AWS customers to fine-tune them for specific use cases, though the underlying training data isn’t visible like in fully open-source models.
• OpenAI positions these models between GPT-3 and GPT-4 in terms of intelligence level.
• The models are text-based rather than multi-modal, making them well-suited for agentic workflows like web browsing and software navigation.

Why this matters: This development erodes Microsoft’s exclusive relationship with OpenAI while expanding enterprise AI choices for AWS customers.

• AWS customers can now access OpenAI models through existing AWS infrastructure without needing separate commercial agreements with OpenAI.
• The move addresses a key gap that AWS customers have been requesting, as Shaown Nandi, who leads AWS’s technical deal team worldwide, noted: “‘You’re missing OpenAI’—that’s what they would say.”
• It provides “air cover” for executives making AI decisions, allowing them to tell boards they’re working with the biggest cloud provider using the biggest variety of open models.

Key performance claims: AWS touts significant cost advantages for the new models compared to competitors.

• The company claims the models are “10x more price-performant than the comparable Gemini model, 18x more than DeepSeek-R1 and 7x more than the comparable OpenAI o4 model.”
• These performance metrics await real-world validation as enterprise customers begin implementation.

Strategic benefits for AWS customers: The integration eliminates operational friction while maintaining existing workflows.

• Customers avoid vendor onboarding processes and can access OpenAI models through familiar AWS tools.
• AI developers don’t need to learn new platforms or rebuild applications to work with the models.
• The models join existing open-weight options from Meta, Mistral, and other providers already supported on Bedrock.

The competitive landscape: This move reflects broader shifts in the enterprise AI market as exclusivity arrangements face pressure.

• Microsoft maintains a complex relationship with OpenAI and still offers a wider range of OpenAI models, including GPT-5 variants, through Azure.
• OpenAI benefits by reaching enterprise customers already tooled for AWS rather than Azure, potentially expanding its ecosystem.
• As Nandi explained, enterprises want different models for different use cases: “They’re not looking for one general-purpose model for agents.”

What industry experts think: The development signals potential changes in traditional AI partnership dynamics.

• Patrick Moorhead, who leads Moor Insights & Strategy, wonders whether this could be “the first chink in the Microsoft–OpenAI armor.”
• The timing coincides with rapid market changes, including disruptive events like DeepSeek’s debut forcing model creators to reconsider training approaches.
• Enterprise customers consistently seek optionality in their AI strategies, making AWS’s broad model selection increasingly valuable.