GPT-5.4 Mini and Nano Now Available

OpenAI's latest compact models—GPT-5.4 Mini and GPT-5.4 Nano—are now accessible through Vercel AI Gateway, expanding the range of models available for developers building AI applications.

GPT-5.4 Mini

GPT-5.4 Mini is designed for agentic tasks requiring reliable code generation, tool orchestration, and multi-step browser interactions. The model balances capability and cost efficiency, making it a strong default choice for complex workflows. Access it by setting the model identifier to openai/gpt-5.4-mini in the AI SDK.

const result = streamText({
  model: 'openai/gpt-5.4-mini',
  prompt: 'Scaffold a new Next.js API route...',
});

GPT-5.4 Nano

GPT-5.4 Nano delivers near-Mini performance at a lower price point, making it ideal for high-volume and cost-sensitive workflows. It excels in scenarios with parallel sub-agent calls where costs scale with volume. Use it with openai/gpt-5.4-nano.

const result = streamText({
  model: 'openai/gpt-5.4-nano',
  prompt: 'Check each file in the PR diff...',
});

Key Features

Both models support verbosity and reasoning level parameters, giving developers control over response detail and computational reasoning depth. These features are particularly valuable for sub-agent workflows where multiple smaller models coordinate on distinct parts of a larger task.

AI Gateway Integration

These models integrate seamlessly with Vercel AI Gateway's existing infrastructure, including unified API calls, usage tracking, cost monitoring, intelligent provider routing with automatic retries, and built-in observability. The gateway also supports Bring Your Own Key (BYOK) for providers.

Developers can explore model performance via the AI Gateway model leaderboard or test models interactively in the playground.