GPT-5.4 mini

GPT-5.4 mini arrived on 17 March 2026, twelve days after the flagship, together with GPT-5.4 nano. OpenAI describes it as its strongest mini model to date for coding, computer use and subagent work — the pattern where a cheaper model executes steps planned by a more expensive one. The context window is 400,000 tokens with 128,000 on output; notably it does not inherit the 1,050,000-token window of GPT-5.4, so the million-token figure belongs to the flagship and Pro tiers only. Input is text and images, the knowledge cutoff is 31 August 2025, and reasoning tokens are supported. Pricing is $0.75 per million input tokens, $0.075 cached, and $4.50 per million output — roughly a third of base GPT-5.4 across the board, and without the 272,000-token surcharge that complicates the larger models, because the window never reaches that threshold in the first place.

▸Videos

No videos yet.

▸Reviews