OpenAI Phases Out o3 and GPT-4.5 From ChatGPT, Triggering User Backlash
OpenAI will remove GPT-4.5 from ChatGPT on June 27 and the o3 model on August 26, the company confirmed in a post on its OpenAI Help Center.
Both models were exclusive to paid subscribers. The company cited limited usage among that group as the primary reason for the retirement, saying it would redirect resources toward its newer GPT-5 series.
The changes affect only direct ChatGPT usage. API access to both models remains unaffected.
What Changes, and When
GPT-4.5 carries a 30-day sunset window, running from May 28 to June 27. O3 gets a 90-day window, remaining available from May 28 through August 26.
Alongside the retirements, OpenAI announced a new version of GPT-5.5 Instant. The company said responses from the updated model read more naturally, feature better pacing, and avoid overly long answers.
OpenAI also confirmed it will remove the Canvas feature — a dedicated writing and coding workspace — from the GPT-5.5 model series. Writing and coding tools will instead integrate directly into standard chat.
Paying users retain Canvas access until o3 and GPT-4.5 reach their respective end dates. Free-tier users will not have that option.
Users Push Back
The announcement drew sharp criticism across Reddit and X, where users questioned the logic of retiring models they were actively paying to access.
Many argued that low usage figures do not simply reflect lack of demand. Rate limits — which restrict how often users can call a given model within a set period — could artificially suppress usage statistics, they said, making light traffic an unreliable proxy for user interest.
Others pointed to a more structural problem: both models sat behind a paywall, limiting the pool of users who could access them in the first place.
Beyond the numbers, users described losing something harder to quantify. Multiple users called o3 distinctive for its conversational tone and natural language flow, arguing that GPT-5 variants do not replicate those qualities. GPT-4.5 drew praise as a writing tool specifically because it resisted overcomplicating simple phrasing — a trait users said newer models lack.
The backlash reflects a broader frustration with OpenAI’s pace of iteration. Critics say the company prioritizes releasing new systems over preserving models that users have grown to rely on. Some framed the retirements as part of a wider drift toward corporate-safe, personality-stripped AI outputs.
OpenAI has not publicly addressed those specific complaints.
Speculation About What Comes Next
An OpenAI employee recently said the company is “making space for what’s coming,” a comment that triggered widespread speculation about an imminent GPT-5.6 release. The company has not confirmed any such announcement.
This is not the first time OpenAI has faced organized user opposition over model retirements. Earlier in 2025, the Keep4o movement gained traction when users protested the planned phaseout of GPT-4o. Around the same time, a fake screenshot falsely claimed OpenAI was offering permanent access through a nonexistent “GPT-4o Legacy Program” — a scam that circulated widely before users flagged it publicly.
