OpenAI retires GPT Nano fine-tuning — users must switch to newer models or lose custom training

OpenAI is shutting down the ability to fine-tune (customize) its GPT Nano model, the smallest and cheapest version of its AI. Users who built custom versions of GPT Nano for their specific tasks will no longer be able to create or maintain them after the deprecation date.

Fine-tuning lets companies train an AI on their own data—customer service responses, legal documents, coding patterns—so it performs better on their specific work. Losing this for Nano means teams either migrate to larger, more expensive models (GPT 4o or GPT 4 Turbo) or stop using OpenAI's fine-tuning altogether.

For business users on tight budgets or working with simple, repetitive tasks, this forces a hard choice: pay more for a bigger model you may not need, switch to a competitor's fine-tuning service, or redesign workflows around OpenAI's newer non-tunable models. Small teams and cost-conscious departments will feel this most.