Amazon Web Services announced that two new versions of OpenAI’s language model are now generally available on its Bedrock platform. The models are called GPT‑6.2 Astra and GPT‑6.2 Sol. They can be accessed through the Bedrock console or via API calls, just like the earlier models that AWS hosts.
Astra is marketed for work that needs deep reasoning or judgment. The company says it can handle complex business questions, research tasks, and advanced coding projects. It also offers an “Ultrafast” mode that reduces response latency for interactive applications such as coding assistants or real‑time agents.
Sol is aimed at high‑volume, repeatable tasks where speed and cost matter. Typical use cases include batch code debugging, document analysis, and multistep workflow automation. Both models run under AWS’s security and governance controls, so users can audit who calls the model and what data is processed.
How the new models differ
The key distinction between Astra and Sol is the trade‑off between performance and price. Astra’s Ultrafast mode is designed for low‑latency interactions, but it may cost more per token than Sol, which is optimized for cheap, high‑throughput processing. AWS does not publish exact pricing in the announcement, directing customers to the Bedrock documentation for regional rates.
Both models support text and image inputs, and they can be integrated into existing AWS services such as Lambda, SageMaker, and Step Functions. This makes it easier for developers to build end‑to‑end AI pipelines without moving data outside of AWS.
What this means for developers
For teams already on AWS, the addition of GPT‑6.2 models removes the need to set up separate OpenAI accounts or manage separate cloud contracts. They can now call the newest OpenAI models from the same console they use for storage, compute, and analytics.
The Ultrafast option gives developers a way to build conversational tools that feel snappy to users, while Sol lets businesses automate repetitive coding or data‑entry tasks at a lower cost. Because the models are hosted by AWS, organizations also benefit from built‑in logging and access‑control features that help meet compliance requirements.
Why it matters
Developers can now use the latest OpenAI models without leaving the AWS ecosystem, simplifying billing and security management. The Ultrafast mode lets customer‑facing apps respond faster, while Sol offers a cheaper option for large‑scale automation. Companies still need to check the detailed pricing and test whether the models meet their accuracy needs.