Back in August 2025, when OpenAI unveiled GPT-5, it also introduced GPT-5 Mini and GPT-5 Nano, two smaller language models designed for high-volume workloads. Since then, OpenAI has rolled out several frontier models, including GPT-5.1, GPT-5.3, and GPT-5.4, but it had not released any new smaller models. This gap allowed competitors like Google to capitalize on the segment with capable mini models such as Gemini 3 Flash.
Today, OpenAI introduced GPT-5.4 Mini and GPT-5.4 Nano, its most capable small models yet, specifically optimized for high-volume workloads. For example, the new GPT‑5.4 Mini performs significantly better in coding, reasoning, multimodal understanding, and tool use when compared to GPT-5 Mini, while running more than 2x faster. In fact, in some popular AI benchmarks, including SWE-Bench Pro and OSWorld-Verified—GPT-5.4 Mini nearly meets the performance of the much larger GPT-5.4 model.
GPT‑5.4 Nano is the most affordable model in the GPT‑5.4 series and performs significantly better than GPT‑5 Nano. OpenAI claims that this model is ideal for generic tasks like classification, data extraction, ranking, and coding subagents.
You can take a look at the benchmark comparisons below:
The new GPT-5.4 Mini model is available in ChatGPT, Codex, and via API for developers. Similar to the larger GPT-5.4 model, the GPT-5.4 Mini supports text and image inputs, tool use, function calling, web search, file search, computer use, and skills. It features a 400k context window and costs $0.75 per 1M input tokens and $4.50 per 1M output tokens. For comparison, the GPT-5 Mini model was priced at $0.25 per million input tokens and $2 per million output tokens.
For developers using the Codex app, CLI, IDE extension, and web, GPT-5.4 Mini will only consume 30% of the GPT-5.4 quota, making it suitable for less reasoning-intensive work. For general consumers, GPT‑5.4 Mini is available inside ChatGPT for Free and Go users via the “Thinking” feature in the + menu. For premium plan users, GPT-5.4 Mini will act as a fallback for GPT‑5.4 Thinking.
OpenAI highlighted that the new GPT‑5.4 Nano model will not be available via ChatGPT or Codex. Instead, it will be available exclusively through the API, priced at $0.20 per 1M input tokens and $1.25 per 1M output tokens.