GPT‑5.4 Mini and Nano

OpenAI just dropped GPT‑5.4 Mini and Nano—smaller, faster versions meant to blitz through everyday tasks—and the internet immediately split into two camps: the speed‑stans and the price‑pain brigade. Fans love the pitch: snappier coding helpers, quick screenshot‑reading tools, and “little robot sidekicks” that execute chores while a bigger model plans the job. One commenter cheered that the old Mini already handled “simple” tasks 99% of the time and hopes this update pushes it closer to perfection.

But the bill? That’s where the fireworks start. Commenters tallied the new prices—Mini at $0.75 input/$4.50 output and Nano at $0.20/$1.25—and yelled “mini in name, maxi on the bill.” Some are asking if earlier cheap models were just loss leaders, while others point out that benchmarks show real gains but not real savings. Another sore spot: Nano isn’t in Codex, the company’s coding suite, prompting “why not for log‑crunching and mass data munching?” Meanwhile, the open‑source crowd rolled in like thunder: one voice said they’ll only be impressed when weights are released (aka, the guts of the model made public).

Between the cheers for faster turnarounds and the groans over higher costs, the vibe is clear: people want speed, but not if it means paying premium rates forever. The big‑brain‑plus‑minions setup (large model plans, small models execute) sounds cool in the docs, but the crowd wants proof it saves money—not just milliseconds.