OpenAI opens its 28-day Codex pledge by speeding GPT-6 Astra and Sol to about 50 tokens a second
OpenAI's Thibault Sottiaux says GPT-6 Astra and GPT-6.1 Sol now stream about 50 tokens a second, up from 30, as day one of a 28-day ship-or-reset pledge. The figure is an approximate streaming rate, so agent run times improve only as far as text generation dominates each loop.
The Engineer · Build desk
What happened
- Subscribers using Sign in with ChatGPT get the faster models inside third-party agents including OpenCode, Pi, Amp and Devin.
- Two days before the pledge, Sottiaux apologized for GPT-6.1 Sol's slow start, blamed a load spike, and reset usage for paid ChatGPT accounts.
- From October 30, the $200 Pro plan's Codex and ChatGPT Work allowance falls from 20 times the Plus allowance to 10 times.
Why it matters
- What it limits constraint Every agent connected through Sign in with ChatGPT draws on one plan allowance, so faster streaming changes how quickly work arrives without enlarging how much work the plan pays for.
- Who has to decide decision Teams choosing a third-party agent on a ChatGPT plan have to time whole tasks on their own code, because the published figure covers text streaming and not tool calls or test runs.
- What it costs cost Pro subscribers keep paying $200 for half the multiplier after October 30, so heavy agent users face an upgrade decision inside the pledge window.
- What it sets up precedent By counting a default speed change as its first "clear improvement", OpenAI has set the bar the remaining 27 days must clear before a reset is owed.
Next story · Build · Oct 8
OpenAI sells the same GPT-6.1 Sol model faster for six times the token priceConfirmed2 publishers
Tokens per second is a streaming rate. A coding agent also plans, calls tools, runs code and checks its work, and output speed does not measure that full cycle, as RuntimeWire pointed out [10]. At the quoted rates, a fixed block of output now streams in about 60% of the time it used to [22]. That cut applies only to the generation share of each loop. A step that waits on a slow test suite gains little.
For the 50 tokens a second to transfer to a given team, the agent's loop has to be bound by generation, and the rate has to hold on that team's product, plan and task mix. Sottiaux called both rates approximate [5]. He also said the models use a highly optimized tokenizer and need fewer tokens per task [6]. OpenAI did not say how it measured the rates, whether they hold across products, plans and tasks, how many tokens the tokenizer saves, or whether it changed usage limits, the models or per-request compute [5][6][7]. The announcement undersells its own figures. The "about 50%" OpenAI put on the change [1] is smaller than the roughly 67% that 30 to 50 implies [21].
The delivery path is good engineering. Sign in with ChatGPT lets an eligible user run a participating app on their ChatGPT plan, with usage counted against the plan and no API key [11]. The speedup shipped as a default, so it reached those apps with no setting to change [4]. Sottiaux said users should feel it within two hours of his October 5 thread on X [9]. The allowance is shared the same way. Users can set weekly limits per app, but an app's limit does not create a separate pool of usage [12].
Resets work as the penalty because OpenAI has made them routine. Sottiaux reset Codex limits for all paid plans in April. He reset usage twice in one day in July after multi-agent regressions, and again when Codex and ChatGPT Work passed 8 million active users. He and Sam Altman handed out a banked reset on stage at DevDay on September 29 [17]. An unofficial tracker had a 28-day scoreboard up within hours of the pledge [18]. One developer quoted by The New Stack suggested burning the allowance every day and hoping OpenAI misses a ship [19]. Mark Kretschmann, a software engineer at ASYS Group, read the pledge as 28 Codex resets in 28 days [20]. "Let the improvements begin," Sottiaux wrote [16].
OpenAI announced the Pro cut alongside a new $500 plan. That plan offers 25 times the Plus allowance and an Ultrafast tier for GPT-6 Astra at up to eight times standard speed in Codex [15]. I'd expect that tier to matter more to latency-sensitive teams than day one did, if the eight times holds on their workloads. For a team choosing among agents on a ChatGPT plan before the window closes, I'd measure task completion time on its own repositories, and how fast each tool drains the shared allowance.
What to watch
- Whether any day in the 28-day window misses a ship and triggers the full usage reset Sottiaux promised.
- Independent task-completion timings for GPT-6 Astra and GPT-6.1 Sol inside OpenCode, Amp or Devin at the new default rate.
- Whether the Ultrafast Astra tier's up-to-eight-times speed holds in Codex once the $500 plan and the October 30 Pro cut take effect.
Clarity's read
What the record supports and how the coverage leans. The claims behind it follow.
Reality
- Evidence35
- Adoption50
- Hype gap+15
- Incentives70
- Confidence60
Claim ledger
Ranked by verification strength, evidence, and original report placement.
- [1]
OpenAI says it increased the default response speed of GPT-6 Astra and GPT-6.1 Sol by about 50% for subscribers using its products and partner apps.
- [2]
Sottiaux said the two models can now generate approximately 50 tokens per second, up from about 30.
- [3]
Subscribers using Sign in with ChatGPT also get the speedup through OpenCode, Pi, Amp and Devin.
- [4]
The change reaches products using Sign in with ChatGPT without users changing their settings.
- [5]
Sottiaux described the rates as approximate and did not specify how OpenAI measured them or whether they apply uniformly across products, plans, and tasks.
- [6]
Sottiaux said the models use a highly optimized tokenizer and need fewer tokens to complete tasks, but gave no token-consumption figure or benchmark alongside that claim.
- [7]
The announcement does not say whether OpenAI adjusted usage limits, changed the models themselves, or altered the underlying compute allocated to each request.
- [8]
Sottiaux 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.
- [9]
Sottiaux wrote in a thread on X on October 5 that the rollout should be felt within two hours of his post.
- [10]
A coding agent can spend time planning, calling tools, running code, and checking its work; token output speed alone does not measure that full cycle.
- [11]
OpenAI describes Sign in with ChatGPT as a way for eligible users to apply their ChatGPT plan to AI requests in participating apps, with usage counted against their plan rather than requiring an API key.
- [12]
Users can set weekly limits for individual apps, and an app's limit does not create a separate pool of usage.
- [13]
Two days before the challenge began, Sottiaux apologized for a slow start with GPT-6.1 Sol, blamed a major load spike in the model's first two days, and announced a global reset for paid ChatGPT accounts.
- [14]
Starting October 30, OpenAI will cut the Codex and ChatGPT Work allowance on its $200 Pro plan from 20 times the Plus allowance to 10 times.
- [15]
OpenAI paired the Pro 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.
- [16]
"Let the improvements begin,"
- [17]
Sottiaux reset Codex rate limits for all paid plans in April; in July he reset usage twice in one day after the ChatGPT desktop app launch introduced regressions in some multi-agent workflows, and reset again when Codex and ChatGPT Work passed 8 million active users; he and Sam Altman handed out a banked reset live on stage at DevDay on September 29.
- [18]
Unofficial services track resets, with one putting up a 28-day scoreboard within hours of Sottiaux's announcement.
- [19]
One developer joked that Sottiaux had turned Codex usage into gambling, suggesting subscribers burn through their allowance each day and hope OpenAI misses a ship.
- [20]
Mark Kretschmann, software engineer at ASYS Group, treated Sottiaux's pledge as 28 Codex resets in 28 days.
- [21]
Going from about 30 to about 50 tokens per second is roughly a 67% increase, not 50%.
- [22]
At about 50 tokens per second instead of about 30, a fixed amount of output streams in about 60% of the previous time, a 40% cut in streaming time.
- [23]
The Pro plan's allowance multiplier halves on October 30, from 20 times Plus to 10 times, at the same $200 price.
- [24]
After the first day's ship, 27 days of the 28-day pledge remain.
Sources
2 independent publishers whose own reporting we read for this story.
- runtimewire.comOpenAI says GPT-6 Astra and Sol now stream 50% faster
1 article · October 5, 2026
- thenewstack.ioDevelopers are secretly hoping OpenAI fails to ship this month
1 article · October 5, 2026
Topics and entities
Topics
- AI Coding Agents
- AI subscription usage limits
- LLM inference speed