GPT-5.4 Mini and Nano Announced: Faster AI Models With Near Flagship Performance

OpenAI has officially introduced GPT-5.4 Mini and GPT-5.4 Nano, expanding its latest AI model lineup with smaller, faster, and more cost-efficient options. The release follows the debut of GPT-5.4 and focuses on delivering strong performance while reducing compute requirements.

These new models aim to make advanced AI capabilities more accessible across apps, APIs, and developer tools.

GPT-5.4 Mini Focuses on Speed and Capability

GPT-5.4 Mini delivers major upgrades in coding, reasoning, multimodal understanding, and tool usage. Despite its smaller size, it achieves performance levels close to the full GPT-5.4 model in several benchmarks.

In tests like SWE-Bench Pro and OSWorld-Verified, GPT-5.4 Mini nearly matches GPT-5.4 while running more than twice as fast as GPT-5 Mini. This positions it as a strong option for developers who need both speed and reliability.

The model supports text and image inputs, along with advanced capabilities such as function calling, web search, file search, and computer use. It also includes a large 400,000-token context window, allowing it to process extensive data in a single session.

GPT-5.4 Mini is available across ChatGPT, Codex, and the API. Microsoft has already integrated both Mini and Nano models into its AI Foundry platform, signaling rapid enterprise adoption.

Pricing and Availability Adjust for Performance Gains

OpenAI has increased pricing compared to the previous Mini model, reflecting the performance improvements. GPT-5.4 Mini costs $0.75 per 1 million input tokens and $4.50 per 1 million output tokens, up from $0.25 and $2 respectively for GPT-5 Mini.

The model also introduces efficiency benefits in developer workflows. In Codex tools, GPT-5.4 Mini consumes only 30% of the GPT-5.4 quota, making it a more economical fallback option.

ChatGPT users can access GPT-5.4 Mini through the “Thinking” feature on Free and Go plans. For premium users, it serves as a fallback when GPT-5.4 Thinking reaches limits.

GPT-5.4 Nano Targets Lightweight AI Tasks

Alongside Mini, OpenAI launched GPT-5.4 Nano as its most affordable model in the series. The Nano variant focuses on lightweight, high-frequency tasks such as classification, data extraction, ranking, and coding subagents.

It delivers noticeable improvements over GPT-5 Nano while maintaining low operational costs. GPT-5.4 Nano is priced at $0.20 per 1 million input tokens and $1.25 per 1 million output tokens.

Unlike Mini, Nano is not available in ChatGPT or Codex. Developers can access it exclusively through the API, positioning it as a backend solution for scalable AI systems.

OpenAI Pushes Broader AI Strategy

The launch of GPT-5.4 Mini and Nano highlights OpenAI’s strategy to diversify its model ecosystem, offering options that balance performance, cost, and speed for different use cases.

In parallel, the company continues expanding its infrastructure footprint. In related developments, OpenAI has reportedly struck a deal with AWS to support AI workloads for U.S. defense and government applications.

Together, these moves signal a push toward wider enterprise adoption, while maintaining accessibility for everyday users and developers.

Via Neowin