BREAKING: OpenAI just dropped GPT-5.4 mini and nano

OpenAI has unveiled GPT-5.4 Mini and GPT-5.4 Nano, positioning them as its most capable small models to date. The star of the show is GPT-5.4 Mini, which delivers performance strikingly close to its larger sibling. It achieves a 54.4% score on the challenging SWE-bench Pro benchmark for coding tasks, just 3.3 percentage points behind the full GPT-5.4 model's 57.7%. This marks a significant leap in reasoning and coding ability for a compact model. OpenAI is making Mini immediately accessible by rolling it out to free and paid ChatGPT users through a new 'thinking' option within the chat interface, democratizing access to advanced, cost-efficient AI.

For developers, the API-only GPT-5.4 Nano serves a different purpose. Priced aggressively at $0.20 per million input tokens, it is engineered for high-volume, low-latency operations like data classification and extraction. OpenAI explicitly targets developers building AI agent systems, where Nano can act as a scalable, affordable workhorse for delegated tasks. Both models represent a strategic push into the 'small but mighty' AI segment, offering substantial improvements in multimodal understanding and tool use over their predecessors. This release allows OpenAI to cater to a broader range of use cases, from free user experimentation in ChatGPT to enterprise-scale agentic workflows, all while optimizing for speed and cost.