[Release Notes] The Most Powerful Small Models Ever: OpenAI GPT-5.4 mini·nano Released

GPT-5.4 mini and nano: mini is available via ChatGPT, API, and Codex; nano is API-only.

Pricing — GPT-5.4 mini: $0.75 per 1M input tokens, $4.50 per 1M output tokens. GPT-5.4 nano: $0.20 per 1M input, $1.25 per 1M output.

Performance — mini is more than twice as fast as the previous GPT-5 mini; SWE‑Bench Pro: mini 54.4% (previous mini 45.7%). Full GPT‑5.4 scored near 57.7%.

Evaluation & nano role — OSWorld‑Verified: mini 72.1% (flagship 75.0, human 72.4). Nano: SWE‑Bench Pro 52.4% (below mini but improved over prior nano-class), OSWorld 39.0%. Nano is optimized for classification, data extraction, ranking, and simple subagents where speed and cost are priorities.

Codex & access — In Codex, mini consumes 30% of the quota compared with GPT‑5.4; full GPT‑5.4 handles planning, coordination, and final judgment. Mini subagents run parallel codebase search, file review, and document processing and handle low‑latency targeted edits, codebase exploration, frontend generation, and debugging loops. ChatGPT Free/Go users can access mini in the Thinking menu; paid subscribers are automatically switched to mini on usage overages.