On September 22, 2026, Anthropic and OpenAI both shipped major model launches — Claude Opus 5.5 and GPT-6 Sol and Luna — roughly two hours apart. Both companies bundled the same specific perk into their announcements: a "banked reset," a usage-limit reset subscribers can save and trigger on their own schedule instead of waiting for the automatic one. The overlap turned Anthropic's own launch-announcement replies into a running joke rather than a serious debate, and it's worth unpacking both why that happened and what real subscriber frustration it's papering over.
TL;DR
| Question | Answer |
|---|---|
| What actually happened? | Both companies shipped the identical "bank a reset, use it whenever" mechanic on the same launch day |
| Who went first? | OpenAI's Tibo Sottiaux posted roughly 90 minutes before Anthropic's Claude account |
| What was the reaction? | Mostly comedy — "who are you and what have you done with the real Claude," jokes about Anthropic hiring Tibo |
| Was it actually copying? | Unresolvable from outside — no evidence either way, just suspicious timing |
| What's the real complaint underneath? | Both platforms run opaque 5-hour rolling windows most subscribers can't predict or reason about |
| Does the reset delay the normal weekly reset? | Unclear on both platforms — asked directly in replies to both companies, unanswered by either as of publication |
What actually happened, in order
OpenAI's Codex lead Tibo Sottiaux posted the GPT-6 Sol and Luna announcement, closing with: "one more thing. We are loading a banked reset into all accounts of our Plus, Pro and Business users." Roughly 90 minutes later, the official Claude account posted its own Opus 5.5 announcement thread, ending on an identical structure: "One more thing: we're increasing five-hour usage limits on Pro, Max, and Team plans. We're also providing subscription users a rate limit reset, which you can save and use whenever you choose." Same mechanic, same "one more thing" framing, same day, within two hours.
Neither company referenced the other. Neither claimed originality. The coincidence did the work on its own.
The reaction was comedy, not accusation
The replies to Claude's announcement read less like a controversy and more like a bit the whole timeline was in on together. Baby Vegeta: "Who are you and what have you done with the real Claude." Nick Iglehart, more pointedly: "A BANKED RESET??? Who are you and what did you do with the real Claude /s" — the trailing /s doing the work of signaling this wasn't a real accusation. Justin.md asked the direct version outright: "Did you guys hire Tibo?!?!?" None of these are serious claims that Anthropic lifted the idea from OpenAI. They're the internet noticing a suspiciously well-timed coincidence and choosing to play with it rather than litigate it.
A separate exchange on X captured the same dynamic from OpenAI's side, one day earlier. When CDAIO posted "Me waiting for Tuesday RESET" in apparent anticipation, Repojournal replied with something closer to actual pushback: "they dropped Opus 5.5 today and didn't give us a reset. I wonder if they know the Tibo Standard for releases?" — before adding, "we are just having fun with these reset tweets cus I seen some of y'all talking nonsense about it. I pay for my own models!" That last line is the tell: even the people making the joke were aware it could read as more serious than intended, and went out of their way to clarify it wasn't.
The genuine questions buried under the jokes
Not every reply was a bit. Several were straightforward, unanswered subscriber questions that both companies' launch posts left open:
Vandanato Anthropic: "Do Cowork users also get a 5-hr session limit increase of 20%? Please be clear. Is it only for Claude Code or even for Cowork / Design users?"Adventure Protectionto OpenAI, weeks after the reset went live: "Does anyone know if using the saved reset will push out the standard weekly reset? i.e. normal reset... will my normal reset be delayed 7 days or will it arrive on the regular schedule?"Tom Duncalfto Anthropic: "If I sign up for Max now, do I get the banked reset? Or only for existing subscribers?"
None of these got a public answer from either company as of publication. That's the actual story under the joke: usage-limit mechanics on both platforms remain genuinely opaque even to engaged, paying subscribers who read the launch announcements closely enough to ask specific, answerable-in-principle questions.
Why this keeps happening
This isn't the first time usage limits have become a flashpoint independent of model quality in 2026. explainx.ai's earlier coverage of Claude Code's usage limit cuts documented a wave of subscriber cancellations driven by opaque changes to five-hour windows, and GPT-6 Sol's own Reddit reaction thread on Codex usage limits tracked the same underlying pattern from OpenAI's side — subscribers repeatedly discover the actual usable ceiling of a "5x" or "20x" plan only by hitting it, not from any published number. Neither company publishes a token-equivalent figure for what these plan tiers actually guarantee, which means every launch-day perk gets evaluated against vibes and prior burn rate rather than any figure a subscriber could plan around in advance. That's the structural reason a genuinely minor perk — a one-time bonus reset — generated this much discussion on both sides: it's one of the only concrete, comparable numbers either company puts in writing about usage limits all year.
There's also a simpler explanation for the same-day timing that has nothing to do with either company copying the other: both labs have been launching within hours of each other repeatedly throughout 2026, a pattern Hacker News commenters flagged directly on the GPT-6 Sol launch thread — "beardsciences: There's no way this wasn't meant to coincide with Anthropic's release today," countered by "jstummbillig: They hinted this release last week for Tuesday already, so if anything it would be Anthropic that tried to make this happen." Nobody outside either company actually knows who scheduled around whom, and the banked-reset overlap reads more like two companies converging on the same obvious lever — a bonus reset is cheap to grant and immediately felt by every subscriber — than evidence of direct copying.
What the launch-day content push looked like underneath the joke
The banked-reset perk didn't arrive alone — both companies backed their announcements with a wave of practical and showcase content the same day, which is worth noting because it shapes how seriously to take the "who copied whom" framing. Anthropic's developer account posted a full Opus 5.5 prompting playbook from Addy Osmani the same day, with concrete advice like dropping "think carefully" from prompts (Opus 5.5 always reasons before replying now) and handing over whole tasks with a defined "done" state rather than micromanaging steps. The official Claude account followed with a thread of creative demos — a generated short story, a recreation of Apollo 8's Earthrise photograph from public data, and a pencil-sketch physics toy brought to life — the kind of showcase content that reads as confidence in the model, not urgency to match a competitor.
That pattern cuts against the "Anthropic panicked and copied OpenAI" reading a little further: a team scrambling to match a rival's perk in the same afternoon doesn't typically also ship a nine-minute prompting guide and a curated demo reel on the same day. More likely, both the reset and the surrounding content were already scheduled, and the reset's timing overlap is closer to coincidence than either company's actual launch-day story.
Honest limitations
- There is no confirmed evidence Anthropic copied the mechanic from OpenAI — the entire premise of the joke is timing-based suspicion, not a documented source of the idea, and usage-reset perks tied to launches are not exclusive to either company historically.
- Neither company has publicly answered whether a banked reset delays the following scheduled reset, a specific and repeatedly-asked question on both platforms as of publication.
- This post covers public reaction and does not audit either company's actual usage-limit math — both companies' exact token-equivalent usage ceilings remain unpublished, which is the underlying condition that makes stories like this recur.
What this means for builders
If you're on either platform, the banked reset itself is genuinely useful — save it for a week you know will be heavy rather than losing it to an automatic weekly trigger you don't control the timing of. But don't read the jokes as confirmation of a real product battle over usage limits between the two companies; treat it as what it visibly was, a coincidence that Twitter had fun with for a day. The more durable signal from this whole exchange is that both companies' usage-limit systems remain opaque enough that even attentive, paying subscribers can't answer basic questions about their own plan from the public documentation — that's the thing worth actually tracking, not who invented the bonus-reset idea first.
Related on explainx.ai
- Claude Opus 5.5 Launch: Every Benchmark and Reaction — the Anthropic launch this perk was bundled into
- GPT-6 Sol and Luna Launch: 50% Price Cuts and Where They Actually Land — the OpenAI launch this perk was bundled into
- Claude Code Limit Cuts: The Reaction and Cancellation Wave — the earlier usage-limit controversy on Anthropic's side this year
- GPT-6 Sol's Codex Usage Limits: The Reddit Reaction — the parallel controversy on OpenAI's side
Primary sources: Tibo Sottiaux on X, September 22, 2026; Claude on X, September 22, 2026, and public replies to both threads.
This post documents public social-media reaction as of September 23, 2026. Neither company had published clarifying answers to the specific subscriber questions referenced here as of publication.