Developers are secretly hoping OpenAI fails to ship this month

OpenAI’s 28-day Codex shipping sprint has its first result, which means subscribers hoping for a free usage reset will have to wait at least another day.

Thibault Sottiaux, engineering lead for Codex at OpenAI, said on Day 1 that the company had made GPT-6 Astra and GPT-6.1 Sol about 50% faster by default. Speeding things up is the first move in an unusually steep challenge Sottiaux set for his team on Sunday, when he promised to ship a “clear improvement” relevant to most Codex and ChatGPT Work users every day for the next 28 days or ship a full usage reset instead.

“Let the improvements begin,” the Sottiaux wrote.

Faster GPT-6 Astra and Sol

Sottiaux put a more concrete number on the change in a follow-up post, saying the two models can now generate approximately 50 tokens per second, up from about 30. Taken literally, that’s closer to a two-thirds jump than 50%, although both figures appear to be rough estimates rather than benchmark results.

The speedup isn’t limited to OpenAI’s own products, since subscribers using Sign in with ChatGPT also get it through OpenCode, Pi, Amp and Devin.

The timing carries some irony. Two days before the challenge began, Sottiaux apologized for a slow start with GPT-6.1 Sol, blaming a major load spike in the model’s first two days, and announced a global reset for paid ChatGPT accounts to make up for it, so the sprint’s first official ship turned out to be a speed improvement to that same model.

The first ship also gives an early sense of what OpenAI is willing to count. Just before announcing the challenge, Sottiaux said the team was “locking in” on simplification, greater efficiency that translates into more usage, “groundbreaking features,” and new models, adding that “feedback is clear that you all want things to get simpler.”

Just before announcing the challenge, Sottiaux said the team was “locking in” on simplification, greater efficiency that translates into more usage, “groundbreaking features,” and new models, adding that “feedback is clear that you all want things to get simpler.”

Codex resets as currency

Usage works as a currency for Sottiaux’s bet because OpenAI has spent months teaching subscribers to watch for resets. When he reset Codex rate limits for all paid plans in April, Sottiaux opened the announcement by joking that resets shouldn’t be handed out for fun because they cost money, before conceding that the vibes were good.

In July, after the launch that folded Codex into the ChatGPT desktop app introduced regressions in some multi-agent workflows, he reset usage twice in a single day, and he marked Codex and ChatGPT Work passing 8 million active users with yet another. He and OpenAI CEO Sam Altman later handed out a banked reset live on stage at DevDay on September 29.

Resets are common enough now that unofficial services track them, with one putting up a 28-day scoreboard within hours of Sottiaux’s announcement, and developers quickly worked out the obvious gamble.

One joked that Sottiaux had turned Codex usage into gambling, suggesting subscribers burn through their allowance each day and hope OpenAI misses a ship. Others would rather have higher limits in the first place. One Plus subscriber said the five-hour cap already dictates when they can work, while Mark Kretschmann, software engineer at ASYS Group, treated Sottiaux’s pledge as 28 Codex resets in 28 days.

That frustration over limits lands at an awkward time, since OpenAI said last week that starting October 30, near the end of the 28-day window, it will cut the Codex and ChatGPT Work allowance on its $200 Pro plan from 20 times the Plus allowance to 10 times.

That frustration over limits lands at an awkward time, since OpenAI said last week that starting October 30, near the end of the 28-day window, it will cut the Codex and ChatGPT Work allowance on its $200 Pro plan from 20 times the Plus allowance to 10 times. The company paired that change with a new $500 plan offering 25 times the Plus allowance and an Ultrafast tier for GPT-6 Astra that promises up to eight times the standard speed in Codex.

Not everyone is rooting for the reset button. The pseudonymous X account Token Gremlin argued that 28 smaller improvements won’t count for much unless OpenAI ships a model that can beat Claude Opus 5.5. That’s a high bar, but an understandable one, given that Opus 5.5 is already pulling developers its way. T3 Code creator Theo Browne said on October 3 that Anthropic’s newest Opus model had become the first to break 50% of traffic in T3 Code, with “literally half of all prompts” going to Opus.

Who judges Codex improvements

The 28-day challenge is not set in OpenAI terms, just in Sottiaux’s posts. Even the dates are unofficial, with one tracker putting the run from October 5 through November 1 and noting that it’s making its own call on the timeline. Sottiaux hasn’t said which subscription tiers would get a reset either, or what exactly a “full reset” would restore, confusion that OpenAI has caused before.

After the April reset for “ALL paid plans,” one Plus user filed a GitHub issue saying their weekly limit stayed on its normal cycle while Codex offered no way to tell whether the special reset had applied at all. According to the issue, OpenAI Support later clarified that the reset did not cover every quota type for every user.

“Full reset” still leaves plenty unanswered. Codex has shorter rolling windows alongside weekly limits, and those are very different things to get back when you’ve run out of usage.

Codex has shorter rolling windows alongside weekly limits, and those are very different things to get back when you’ve run out of usage.

The post Developers are secretly hoping OpenAI fails to ship this month appeared first on The New Stack.