OpenAI $200 Plan Cut to 10x Usage: What Changes and Costs?

What is changing in OpenAI's and Anthropic's subscription usage limits?

According to a casual discussion among AI creators, OpenAI has launched a $500 tier, and its $200 plan is set to drop from 20x usage to 10x about 30 days after Dev Day. Anyone buying the $200 plan now reportedly gets 10x anyway, so the cut affects existing subscribers. The same discussion says Anthropic is A/B testing its own usage limits. None of this was confirmed by an official source in the material reviewed.

What did the hosts actually say about the $200 plan?

The discussion started with a viewer question: would Anthropic watch OpenAI cut usage in half after October ends and do nothing? One host answered that OpenAI had said, after about a month of the new $500 plan being out, the $200 plan would be cut in half.

Another host sharpened it. The $200 plan is going from 20x to 10x. That applies to current $200 subscribers, because anyone who buys the plan now gets 10x regardless. A third host described the timing as "like 30 days after Dev Day."

The sources do not say what the "x" multiplier is measured against. It most likely refers to a multiple of a baseline paid tier's allowance, but treat that as an assumption. The hosts also gave no token counts, message counts, or rate-limit windows.

What is the $500 tier?

The brief and the conversation describe a new $500 plan from OpenAI that sits above the $200 plan. The hosts gave no feature list, no multiplier for it, and no billing details. Whether $500 is monthly was not stated, though subscription pricing like this is usually quoted monthly.

One host said he used 100% of his $500 quota every day while editing. That tells you the tier is aimed at heavy, sustained use, and that a single power user can burn through it. It does not tell you how large the quota is.

Why would OpenAI cut a plan right after launching a pricier one?

The hosts did not claim to know OpenAI's reasoning. One offered a read on the strategy: announce a cut, then soften it by doubling the $500 tier instead of leaving things as they are. He said he could see OpenAI doing that, and framed it as speculation.

The logic is easy to follow, though. A new top tier gives heavy users somewhere to go. If the $200 plan keeps its old allowance, the $500 plan has less reason to exist. Narrowing the gap between tiers pushes power users upward. Compute is also expensive, and flat-rate plans get stretched by users who run agents and coding tools all day. That is general industry context, not something the hosts said about OpenAI's finances.

What are the usage resets?

A host mentioned that an OpenAI developer-relations figure (referred to as "Tibo") had posted a list of things being dropped, and that the list included a usage reset. Another host said a second reset followed. The one who said he used his full $500 quota daily called the back-to-back resets "so nice."

A reset restores your allowance before the normal window ends. For heavy users, that means more working time without waiting. The hosts did not say how large the resets were or whether more are planned. If the $200 reduction is coming, resets in the meantime soften the transition without changing the long-term math.

Is Anthropic really A/B testing its usage limits?

One host stated flatly that Anthropic is not doing nothing and is "A/B testing their usage right now." That was the full extent of it. The discussion offered no details on:

• which plans are in the test
• whether limits are going up or down for the test group
• how long the test runs
• how users would know which group they are in

A/B testing limits is a common way for a company to find out how users behave under different caps before it commits to a public change. Two accounts on the same plan can see different behavior during a test like this. If your Claude limits feel different from a colleague's, that is a plausible explanation, but it remains unconfirmed.

How should you plan around this?

The practical advice is mundane.

1. Check the official plan page for your tool before you rely on any multiplier you heard in a stream.
2. Track your own usage for a week or two so you know whether 10x versus 20x matters for your workload.
3. Avoid annual commitments if your workflow depends on a specific allowance, since these allowances appear to shift.
4. Spread work across tools. The hosts themselves mix several subscriptions. One named Anthropic and ChatGPT as his main two, plus an X bundle that folds in other products, and said he picks different tools for different jobs: a personal agent for assistant tasks, and Codex and Claude for coding.

That last point is the real hedge. If one provider tightens limits, a second subscription keeps you working.