We’ve just added support for OpenAI’s newest models, GPT-5.4 Mini and GPT-5.4 Nano, to the Vercel AI Gateway. These models are built for speed and efficiency, making them ideal for complex tasks like coding and computer use.
Here is what you can do with them:
- Build sub-agent workflows: These models are designed for tasks where multiple smaller models coordinate to solve a larger problem.
- Fine-tune your responses: You can now use the
verbosityandreasoning_levelparameters to control how much the model reasons before it answers. - Reliable tool orchestration: GPT-5.4 Mini is a strong default for agentic tasks, handling code generation and multi-step browser interactions more reliably than previous versions.
You can start using them today by setting your model to openai/gpt-5.4-mini or openai/gpt-5.4-nano in the AI SDK.
Are you planning to use these for sub-agent workflows? Let us know what you’re building below! 