AI-Starter 3.0.1
We are pleased to announce the release of AI-Starter version 3.0.1.
Key Updates
- Support for new AI models
- Added integration options with AgentCore Runtime (Beta)
- Various bug fixes and minor improvements
1. Support for New AI Models
We have added support for the following AI models:
- OpenAI GPT-5 series
- Azure OpenAI GPT-5 series
- Azure OpenAI Model Router latest version
- Claude Opus 4.1
The characteristics of each GPT-5 series model are as follows. While they currently tend to take longer to generate responses, they provide high-quality answers. The cost tends to be low across the entire series.
- GPT-5: Our flagship model for various domains including coding, reasoning, and agent tasks.
- GPT-5 Mini: A model with excellent balance of speed and cost.
- GPT-5 Nano: A model specialized for ultra-low latency and low cost.
- GPT-5 Chat: The GPT-5 snapshot model currently used in ChatGPT (web app).
Azure OpenAI Model Router is a model for automating model selection in question answering. In this update, we have added support for the 2025-08-07 version. If already enabled, please update the version of the target resource to 2025-08-07 in the Azure OpenAI management console. For models used in this version, please refer to the official documentation:
https://learn.microsoft.com/azure/ai-foundry/openai/concepts/model-router#underlying-models
Claude Opus 4.1 is the latest version of the top-tier model Opus 4. It shows overall performance improvements, with particularly dramatic improvements in coding performance. As before, it tends to be costly and take longer to generate responses, but provides high-quality answers.
https://www.anthropic.com/news/claude-opus-4-1
For a complete list of other currently supported models, please refer to the following guide:
2. Added Integration Options with AgentCore Runtime (Beta)
We have released integration options with AgentCore Runtime (BedrockAgentCoreRuntime).
AgentCore is a new service from AWS, currently available in preview, for securely deploying, running, and scaling AI agents. AgentCore Runtime is a serverless environment that hosts agents.
With this update, we have started offering options to execute AgentCore Runtime from AI-Starter. This option is being offered as a beta version. For details on integration with AgentCore, please contact our support team.
For a complete list of currently supported options, please refer to the following guide:
Conclusion
Thank you for reading to the end. AI-Starter continues to improve based on feedback from our users. Please review this update and share any feedback or requests you may have. Your input drives the evolution of AI-Starter. Stay tuned for future updates.