Skip to content
OpenAI

GPT-5.4 Mini

Mid-size OpenAI tier — solid quality at $0.75/$4.50.

API onlytextimagefileREL 2026-03-17
Compare this model →Verified 2026-06-12

Overview

GPT-5.4 Mini is the balanced small model in OpenAI's lineup: a 400K context window, multimodal input, and enough reasoning ability for most assistant and RAG workloads at $0.75/$4.50 per 1M tokens.

Benchmark scores

Methodology →
Intelligence49
Codingn/a
Reasoningn/a
Mathn/a

Sources and verification

Strengths and weaknesses

Strengths

  • Good cost/quality middle ground
  • 400K context
  • Fast

Trade-offs

  • Noticeable quality gap to full 5.4 on hard tasks

Reach for this model when

RAG answers, app backends, everyday assistant features.

Run something like it locally

A 30–70B open model (Qwen, Llama 4) on Ollama or LM Studio competes directly at this tier.