Anthropic Reset Everyone's Claude Limits. The $900B Valuation Explains Why.

Anthropic Reset Everyone's Claude Limits. The $900B Valuation Explains Why.

4 min readMay 18, 2026

Quick verdict

Anthropic reset all 5-hour and weekly Claude rate limits this week, on top of the doubled-limits announcement that followed the SpaceX compute deal. It is, as Yuchen Jin put it, "running the Codex playbook." Loosen the leash, win developer mindshare back. FT reporting cited by Kimmonismus puts Anthropic's valuation at $900B and ARR at $45B by end of May.

The interesting part isn't the reset. It's that Anthropic is now competing on developer experience, not benchmarks.

What changed today

  • All 5-hour rate limits reset for every user with an active Claude session.
  • Weekly limits reset across Claude Code Pro and Max tiers.
  • Announced directly by @ClaudeDevs, the official Claude developer account.
  • Comes on top of the earlier compute-driven limit increases tied to the Anthropic-SpaceX deal.

If you hit the wall earlier this week, you can pick up where you left off. If you were rationing Claude Code usage, stop rationing for now.

The financial context that makes it make sense

The number that got buried: Anthropic is reportedly at $900B valuation and $45B ARR by end of May, per FT figures cited in the @kimmonismus tweet. That's up sharply from earlier checkpoints. For comparison, OpenAI's last reported numbers were in the same band, give or take. The two companies are now within striking distance of each other on financial scale.

At $45B ARR, you can afford to be generous with rate limits. The math on a single Claude Code Pro user hitting weekly caps is "they pay $200/month either way, give them the headroom." Throttling at this scale costs you developer goodwill and gains you almost nothing in compute savings. The marginal cost of letting someone burn an extra million tokens is rounding error against the revenue line.

For most of Q1 2026, that math went the wrong way. Compute was constrained, capacity was tight, and Anthropic had to ration. The SpaceX deal solved capacity. The financial growth justifies the loosening. This week is the visible result.

Why this is the "Codex playbook"

OpenAI Codex spent late 2025 and early 2026 quietly being more generous than Claude Code. Same nominal price, more actual usage. That generosity is what built the 4 million weekly active users number OpenAI announced with Codex mobile.

Anthropic noticed. The pattern is now:

  1. Get more compute (SpaceX deal).
  2. Loosen the throttles.
  3. Let developers actually use the product they're paying for.
  4. Win back the people who switched to Codex because of capacity, not quality.

It's a defensive move dressed up as a gift. Both can be true at once.

Video: what the reset actually means

What stays uncertain

A reset is not a structural change. The 5-hour and weekly limits still exist. Anthropic just refilled the bucket. If the next two weeks of usage exhaust the new capacity, the throttles come back.

The $900B / $45B numbers are FT-cited but unverified by Anthropic publicly. Take them as directionally correct, not precisely confirmed. Either way, the order of magnitude is clear: Anthropic is now a frontier-scale business, not a research lab with a chat product.

What this doesn't fix: the weekly cap as a structural feature still annoys heavy users. The compute-deal-plus-reset sequence buys six to twelve months of breathing room before the next round of rationing complaints starts.

What to do with this

If you've been using Claude Code carefully because of caps, stop. Run the workflows you were holding back.

If you switched to Codex or Cursor over capacity, re-evaluate. The thing that pushed you out may not exist this week.

If you build on the Anthropic API, the rate limit reset applies there too in spirit. Your 429s should drop. Re-test your retry budgets.

If you're paying for both Claude Code and ChatGPT to hedge against caps, you can probably drop one for a month and see how it goes. Our cost stacking guide covers when that math works.

FAQ

Do the new limits apply to me retroactively?

If you had an active subscription when the reset hit, yes. Your 5-hour and weekly windows started over.

Is this permanent?

The reset is a one-time event. The new ceiling (post-SpaceX-deal capacity) is structural for now. Both can change if demand grows faster than capacity.

Did Anthropic change the per-message cost?

No. Same model, same pricing. Just more headroom inside the existing plans.

Is the $900B valuation confirmed?

FT reporting, cited via Kimmonismus on X. Not officially confirmed by Anthropic. Treat it as directional.

Should I cancel Codex now?

Not on one announcement. Run a real two-week comparison. Codex still has things Claude Code doesn't (mobile, browser plugin). Our three-way breakdown is the place to start.

Sources

Further Reading

Try all the models mentioned in this article

Admix gives you GPT-5, Claude, Gemini, and 350+ AI models in one app. Compare responses side by side. Free to start.

Start free on Admix

Related articles