# Claude Code Doubles Its Rate Limits: What Changed After the SpaceX Deal

> On May 6, 2026 Anthropic doubled Claude Code rate limits for Pro, Max, Team and Enterprise, and removed the peak-hour restriction after a compute deal with SpaceX. Here's what actually changed in practice.

## What Exactly Changed on May 6, 2026

Three main changes:

1. **5-hour rolling limit doubled** in Claude Code for Pro, Max, Team and Enterprise.
2. **Peak-hour throttling removed** on Pro and Max accounts.
3. **«Considerable» increase** in API rate limits for Claude Opus models.

Everything applies immediately. Nothing to update.

## Summary Table: Before vs After

Anthropic has never published exact numbers. Figures come from empirical measurements by the developer community.

| Plan | Before (5 h, Sonnet) | After | Peak hours |
|---|---|---|---|
| Pro · $20/month | ~10-45 messages | ~20-90 messages | No throttling |
| Max 5× · $100/month | 5× Pro | ~10× Pro effective | No throttling |
| Max 20× · $200/month | 200-800 prompts/session | ~400-1,600 prompts/session | No throttling |
| Team / Enterprise | Variable per seat | Doubled per seat | Unchanged |

## How Many Messages Claude Code Pro Gives You Now

$20/month now gets you roughly 20-90 messages per 5-hour rolling window with Sonnet (was 10-45). Turns Pro into a viable option for half-day work sessions.

Two things that didn't change:

- Claude Code and the claude.ai web app **share quota** on Pro and Max.
- The **weekly cap** (introduced August 2025) remains active for heavy users.

## What Changes for Claude Max 5× and Max 20×

- **Max 5×**: net effect equivalent to ~10× the original Pro quota.
- **Max 20×**: sessions of 400-1,600 prompts (was 200-800).

If you were hitting Pro's limit several times a week, the upgrade to Max 5× makes even more sense now.

## Why the End of Peak-Hour Throttling Is the Bigger Story

Before, Anthropic dropped effective limits during periods of highest demand in the United States. For users in Europe, LATAM and Asia, that often coincided with their late-afternoon work window. With the change, those caps disappear.

**Developers in European, LATAM and Asian time zones are the biggest winners.**

## What's Happening With Claude Opus on the API

Anthropic says it raised Opus rate limits «considerably» without publishing figures. Matters for:

- Teams building agents on Opus.
- Developers using Opus for complex reasoning.
- Apps with variable traffic.

If you have retry logic with exponential backoff for Opus, review headers and simplify if the new limits suffice.

## The Anthropic-SpaceX Deal in Numbers

- **Data center**: Colossus 1, owned by SpaceX.
- **GPUs**: 220,000+ NVIDIA.
- **Electrical capacity**: 300+ megawatts.
- **Timeline**: within the first month of the announcement.

300 MW equals the continuous consumption of about 250,000 average homes. Dedicated entirely to training and serving Claude models.

Elon Musk said he decided to lease capacity to Anthropic after being impressed by their work to make Claude AI «good for humanity».

## Why It Matters Beyond the Headline

Anthropic has had the worst capacity-to-demand ratio in AI top three throughout 2025 and early 2026. This translated into:

- Frequent 529 errors (overloaded), especially during US business hours.
- Variable latencies.

With 300 MW of additional capacity in the first month, both should ease. **The subtext: «we're exiting the compute scarcity regime».**

## How to Monitor Your Real Claude Code Usage

1. **`/usage` command** inside Claude Code.
2. **`anthropic-ratelimit-*` headers** if calling the API directly.
3. **Usage panel on claude.ai**.

Good practice: run `/usage` before large tasks. If you've burned 60% of the window, split the task.

## What Didn't Change: the Weekly Cap

The weekly cap introduced August 2025 remains active and **was not doubled**. Works as a parallel ceiling: even with margin in the 5-hour window, exceeding the weekly cap drops effective limits.

For most users it's invisible. For extremely intensive users, the option remains Max 20× or enterprise API usage.

## What to Expect in the Next Few Months

Three things to watch:

- **529 errors**: if they drop noticeably, the investment is working.
- **API pricing**: with more capacity, room for selective price drops on Sonnet.
- **Orbital data centers**: Anthropic mentioned interest in multi-gigawatt orbital compute with SpaceX.

## Next Steps

- If you don't use Claude Code yet: the [main guide](https://claudecodeguia.com/en/) sets you up in 10 minutes.
- Cutting token spend: [LLM caching guide](https://claudecodeguia.com/en/llm-caching/).
- Anthropic design tool: [Claude Design guide](https://claudecodeguia.com/en/claude-design/).

If Pro is tight even after doubling, try Max 5× for a month before jumping to Max 20×.