OpenAI · 2026-03-17 · notable
GPT-5.4 mini and nano
OpenAI's economy-tier models: mini runs 2x faster than GPT-5 mini and approaches GPT-5.4 on SWE-Bench Pro; nano is the cheapest option for classification and extraction. Mini is free-tier; nano is API-only.
OpenAI's smallest GPT-5.4 variants — mini is free-tier fast, nano is API-only cheap, both punch above their weight.
Key specs
| Mini pricing | ~$0.40/M input |
|---|---|
| Speed vs gpt 5 mini | 2x faster |
What is it?
GPT-5.4 mini and GPT-5.4 nano are the economy-tier variants of GPT-5.4, released March 17, 2026 — twelve days after the main launch. Mini is available to free-tier ChatGPT users and significantly improves over GPT-5 mini across coding, reasoning, multimodal understanding, and tool use. Nano is the smallest and cheapest variant, API-only, built for tasks where speed and cost matter most: classification, data extraction, ranking, and coding subagents.
How does it work?
Mini runs more than 2x faster than GPT-5 mini and approaches the full GPT-5.4 on several evaluations including SWE-Bench Pro and OSWorld-Verified. Nano is recommended for high-volume, lower-complexity tasks. Both are designed for the developer sweet spot of 'good enough quality at minimal cost and latency.'
Why does it matter?
Most production AI workloads do not need frontier-tier reasoning — they need fast, cheap, reliable models for classification, extraction, and routing. Mini and nano fill that gap in the GPT-5.4 family, and mini being free-tier means hundreds of millions of ChatGPT users now access GPT-5.4-class capability at no cost.
Who is it for?
High-volume API users, developers building multi-agent systems that need cheap routing models.
Try it
platform.openai.com/docs/models/gpt-5.4