Migrating: GPT-5.4 Pro to GPT-5.5 Pro

A premium-tier-to-premium-tier migration where Standard pricing doesn't move at all — every input and output rate matches exactly between these two, at every tier both publish.

What actually moves

Knowledge cutoff advances to GPT-5.5 Pro's more recent training data — the concrete benefit of this migration, and with Standard pricing unchanged, effectively the primary one. Rate-limit group changes, and it's worth checking directly rather than assuming continuity: the two Pro models publish different limit tables, with different entry-tier RPM and TPM ceilings relative to the same large context window both share.

What stays put

Context window and maximum output length are identical between the two. Neither publishes Fast mode. GPT-5.4 Pro states the long-context repricing rule explicitly; GPT-5.5 Pro's own page doesn't state it either way, which this site records as not published rather than assumed to carry over.

What to check before switching

Since Standard pricing is unchanged, the real work in this migration is re-checking the rate-limit table specifically — a workload's request pattern that fit comfortably under GPT-5.4 Pro's entry-tier ceiling isn't guaranteed to fit under GPT-5.5 Pro's, because the two tables genuinely differ even though the sticker price doesn't. Run the actual request shape through the rate-limit tier planner against both models before assuming the migration is a pure, unconditional upgrade — recency is a real win here, but it isn't the whole story.