Illustration for: OpenAI rolls back GPT-5 router after it lowered daily active users
LLMs & Generative AI

OpenAI rolls back GPT-5 router after it lowered daily active users

2 min read

OpenAI’s latest tweak to its GPT‑5 rollout has sparked a quiet alarm among product teams. After months of touting a new “router” meant to streamline how the model tackles tangled prompts, the company pulled the feature back—citing a dip in daily active users. The move is noteworthy because the router was billed as a direct answer to the very bottleneck that now seems to be hurting engagement: the time it takes for the system’s reasoning engines to churn through intricate queries.

While the idea was to let users glide past complexity, early data suggested the opposite, with sessions stretching into minutes and users slipping away. The rollback underscores a tension between architectural ambition and real‑world usage patterns. It also raises a simple question: when does adding sophistication start to cost a platform its audience?

The answer, according to insiders, lies in the numbers—daily active user counts that fell as the router lingered in production.

*“A router could cut through that complexity — something OpenAI had explicitly aimed for when designing GPT‑5. According to a source familiar with the situation, the router hurt daily active user numbers because reasoning models can take minutes to work through complex questions. Daily active users is…”*

A router could cut through that complexity - something OpenAI had explicitly aimed for when designing GPT-5. According to a source familiar with the situation, the router hurt daily active user numbers because reasoning models can take minutes to work through complex questions. Daily active users is a key success metric for OpenAI.

Routing to better models proved expensive There's another factor at play: routing more queries to expensive models costs OpenAI money, and the AI company is already burning through cash at a staggering rate. Shortly after launch, Altman said the router had boosted reasoning model usage among free users from under one percent to seven percent, and from seven percent to 24 percent for Plus subscribers.

Related Topics: #OpenAI #GPT-5 #router #daily active users #reasoning models #AI #Altman

Did the rollback achieve its goal? OpenAI quietly pulled the GPT‑5 router from free and $5‑tier accounts, forcing those users onto GPT‑5.2 Inst. The router had been marketed as a way to cut through model complexity, automatically matching queries to the most suitable engine.

Yet a source familiar with the change says the feature correlated with a dip in daily active users, apparently because the reasoning models it routed to could take minutes on tough prompts. Faster responses, it seems, were more valuable to the broader user base than deeper reasoning. The move underscores a broader tension: users have been conditioned to equate speed with quality, while AI sometimes rewards patience.

OpenAI’s decision to revert rather than educate users about the trade‑offs leaves an open question about long‑term strategy. A puzzling choice. It is unclear whether the rollback will stabilize engagement metrics or simply mask underlying usability concerns.

The episode highlights that simplifying AI access may require more than technical fixes; user expectations must be managed alongside feature design.

Further Reading

Common Questions Answered

Why did OpenAI roll back the GPT-5 router after its initial rollout?

OpenAI removed the GPT-5 router because it was linked to a decline in daily active users. The company observed that the router sent complex queries to reasoning models that often required minutes to respond, hurting engagement.

How did the GPT-5 router impact daily active user numbers according to the article?

The article states that the presence of the router correlated with a dip in daily active user metrics. Users were less likely to stay active when the router routed them to slower, more resource‑intensive reasoning engines.

What cost concerns arose from routing queries to more expensive models in the GPT-5 system?

Routing more queries to higher‑tier reasoning models increased OpenAI's operational expenses, as these models consume more compute power. The article notes that this financial strain was a key factor in deciding to pull the router feature.

What change was made to free and $5‑tier accounts after the GPT-5 router was removed?

After the rollback, OpenAI forced users on free and $5‑tier plans to use the GPT-5.2 Inst model directly, bypassing the router entirely. This shift aimed to provide faster responses and recover the lost daily active user base.