Amazon Integrates OpenAI Models with AWS, Strengthening AI Catalog and Partnership

Amazon Web Services (AWS) has partnered with OpenAI to add its gpt-oss-120b and gpt-oss-20b models to its AI platforms. Bank of America analysts see this as a positive step in the relationship, strengthening AWS's growing AI catalog and adding revenue-generating services. The models are less advanced than OpenAI's flagship GPT-4, but fit well into AWS's cost savings strategy.

Amazon Web Services (AWS) has expanded its AI capabilities by partnering with OpenAI to integrate its gpt-oss-120b and gpt-oss-20b models into its AI platforms. This strategic move is seen as a positive development by Bank of America analysts, who believe it will enhance AWS's growing AI catalog and generate additional revenue streams. Although these models are less advanced than OpenAI's flagship GPT-4, they align well with AWS's cost-saving strategy.

The gpt-oss-120b and gpt-oss-20b models are open-weight language models designed to deliver strong real-world performance at a low cost. They outperform similarly sized open models on reasoning tasks and demonstrate robust tool use capabilities. The models are optimized for efficient deployment on consumer hardware and are available under the flexible Apache 2.0 license [1].

The gpt-oss-120b model achieves near-parity with OpenAI o4-mini on core reasoning benchmarks and can run efficiently on a single 80 GB GPU. Meanwhile, the gpt-oss-20b model performs comparably to OpenAI o3-mini on common benchmarks and can run on edge devices with just 16 GB of memory, making it ideal for on-device use cases and rapid iteration without costly infrastructure [1].

Both models also exhibit strong performance on tool use, few-shot function calling, and CoT reasoning, as seen in results on the Tau-Bench agentic evaluation suite and HealthBench. They are compatible with AWS's Responses API and are designed to be used within agentic workflows with exceptional instruction following, tool use like web search or Python code execution, and reasoning capabilities [1].

OpenAI's approach to releasing these models emphasizes safety, with comprehensive safety training and evaluations, including an adversarially fine-tuned version of gpt-oss-120b under their Preparedness Framework [1]. The models perform comparably to OpenAI's frontier models on internal safety benchmarks, offering developers the same safety standards as their recent proprietary models.

AWS's integration of these models is expected to provide developers with more options to choose the performance, cost, and latency they need to power AI workflows. The partnership also aligns with AWS's commitment to providing open and flexible AI solutions, empowering both individual developers and large enterprises to run and customize AI on their own infrastructure.

References:
[1] https://openai.com/index/introducing-gpt-oss/