OpenAI launches GPT-5.4 mini and nano: Here's what these AI models can do

OpenAI, has launched two smaller versions of its model, named GPT-5.4 mini and GPT-5.4 nano, with the aim of providing faster response times and efficiency in handling high-volume AI-related operations.

The firm explained that the two models are meant to make AI tools faster and more cost-effective for day-to-day operations while improving coding, reasoning, and multimodal capabilities.

Also Read | ‘Almost human’: Mathematician stunned as AI solves complex math problem designed 20 years ago

GPT-5.4 mini designed for everyday AI tasks

According to OpenAI, GPT-5.4 mini is designed for completing a variety of tasks without requiring computing power from other AI models.

It improves upon the previous GPT-5 mini architecture with enhanced performance for coding, reasoning, and handling multiple modes of input and integrating with tools. OpenAI claims that GPT-5.4 mini is more than twice as fast as the previous version.

The company also noted that GPT-5.4 mini’s performance in benchmarks such as SWE‑Bench Pro and OSWorld‑Verified brings it closer to the capabilities of the full GPT‑5.4 model. The GPT-5.4 mini is also available for use through ChatGPT. Users of ChatGPT can access GPT-5.4 mini through the ‘Thinking’ option available in the menu of ChatGPT.

According to OpenAI, GPT-5.4 mini will be available for free and Go users through this feature. However, for paid users, GPT-5.4 mini will be used automatically when limitations for GPT-5.4 Thinking are reached. GPT-5.4 mini is also available through OpenAI’s API and Codex, their coding tool.

GPT-5.4 nano built for high-volume workloads

The second model, GPT-5.4 nano, is intended for tasks that value speed and cost efficiency over deep reasoning. According to OpenAI, the GPT-5.4 nano model is intended for tasks such as classification, data extraction, ranking systems, and light coding automation.

Also Read | Instagram to remove message encryption from May 8: Here's what it means for users

Due to its small size, GPT-5.4 nano is intended for tasks that require processing numerous requests using fewer computing resources.

This release comes after several updates to OpenAI’s model list, including GPT-5.4 Thinking and GPT-5.3 Instant. The updates were intended for enhanced reasoning and response speed in the company’s AI services.