OpenAI is pushing faster AI with roll out of GPT-5.4 mini and nano models

ChatGPT

• OpenAI releases GPT-5.4 mini and nano for faster, cheaper AI tasks
• GPT-5.4 mini offers improved coding, reasoning, and tool usage
• GPT-5.4 nano targets high-volume, cost-sensitive AI workloads

Did our AI summary help?

OpenAI is expanding its model lineup again, this time with GPT-5.4 mini and GPT-5.4 nano—positioned as its most capable small models so far. The focus is clear: faster responses, lower costs, and better performance for everyday AI workloads.

Here’s how the rollout works.

How GPT-5.4 mini improves everyday AI

GPT-5.4 mini is now live in ChatGPT and is designed to handle a wide range of tasks without the overhead of larger models. It builds on GPT-5 mini with noticeable gains in coding, reasoning, multimodal understanding, and tool usage.

OpenAI says the model runs more than twice as fast as its predecessor, while getting closer to the performance of the full GPT-5.4 model in benchmarks like SWE-Bench Pro and OSWorld-Verified.

In simple terms, it’s meant to feel quicker and smarter without feeling “lite.”

How to use GPT-5.4 mini in ChatGPT

Access depends on your plan, but OpenAI is widening availability:

Free and Go users can access it through the “Thinking” option in the + menu

Paid users will see it as a fallback option for GPT-5.4 Thinking when limits are reached

It’s also available in Codex and OpenAI’s API

This effectively makes a more capable model available to a much broader user base.

How GPT-5.4 nano fits into the picture

At the other end, GPT-5.4 nano is built for scale. It’s the smallest and cheapest version in the lineup, designed for tasks where speed and cost matter more than depth.

OpenAI recommends it for Classification, data extraction, ranking tasks, and lightweight coding agents

Think of it as the background worker—handling repetitive, high-volume tasks without needing the full intelligence stack.

The launch follows GPT-5.4 Thinking, which introduced six upgrades earlier this month, and GPT-5.3 Instant, which focused on faster, more natural responses.