|
OpenAI has put a hard stop on GPT-5.5: October 14, 2026. The announcement is short on ceremony — no migration guide bundled with it, no discount to ease the switch — but the date itself is the whole story for anyone still routing production traffic through that model. GPT-5.5 was the workhorse a lot of solo builders quietly standardized on before GPT-6 Astra existed, and its sunset means every integration built against it needs a new endpoint, a re-tested prompt set, and a fresh look at pricing before the switch flips.
The options aren't neutral. Astra ships with real capability gains but also carries the baggage of a messy rollout — an apology from Altman over access failures, a walked-back AGI claim, and a Pro-tier demand crunch that forced a temporary subscription pause earlier this month. GPT-6-sol, the cheaper sibling OpenAI put on the API alongside Astra, is the more likely landing spot for cost-sensitive workloads, but it hasn't accumulated the same track record yet. Either way, this is exactly the kind of access change this newsletter tracks and generic 'model upgrade' coverage doesn't: a specific date, a forced migration, and a budget decision that a solo operator has to make without much runway.
It also isn't an isolated event. Google set its own hard sunset — October 2, 2026 — for Gemini 2.5 Flash Image, pushing users toward 3.1 Flash Image or Flash Lite Image. Two majors closing out legacy models within two weeks of each other in the same quarter is a pattern, not a coincidence: as frontier releases stack up, the tail of older, cheaper models that builders leaned on for stability is getting trimmed faster than before. If you've been treating a specific model ID as a stable foundation for a side project or client tool, this is the reminder that none of them are permanent — check your dependencies against the sunset calendar, not just the price sheet.
More this week:
|
