Claude Code Slow Today? Here's Why (and What to Do)

If Claude Code feels slow today, the most likely cause is one of three things: you've hit a usage rate limit and responses are being throttled, Anthropic's API is under heavy load, or there's an active incident on their infrastructure. Check the Anthropic status page first. If it's green and you're still getting lag, the problem is almost certainly on the usage/throttle side.

• Claude Code enforces a 5-hour rolling usage window — heavy use slows responses before a hard stop
• Throttling often starts silently: responses get longer to arrive before you see an explicit limit message
• Pro plan users get significantly less throughput than Max plan users; slowness mid-session is a common symptom

Why is Claude Code slow today?

Claude Code's speed depends on three factors: your remaining usage quota, Anthropic's API server load, and your network connection to the API. Quota-related slowdowns are the most frequent culprit. When you're approaching your 5-hour window limit, the API begins throttling responses — requests queue longer before they're processed. This produces the classic symptom of Claude Code "thinking" for much longer than usual without producing output.

Usage throttling (most common cause)

Claude Code bills usage against a rolling 5-hour window. As you approach the cap, response latency increases noticeably before any hard error is shown. If you ran heavy agentic tasks earlier in the session — long file diffs, multi-file edits, or extended conversations — you may be getting throttled even if you haven't seen a "rate limit exceeded" message yet.

You can check your current usage with the /usage slash command inside any Claude Code session, or by visiting claude.ai/settings/usage. Both show how much of your current window has been consumed. See also: when does Claude Code usage reset and how to check your Claude Code usage limits.

Anthropic server load or incident

High-traffic periods (US business hours, major product launches) slow the API for everyone, not just users near their limits. An active incident will be listed at status.anthropic.com. If there's a degraded API or elevated error rate, waiting it out is the only fix.

Network or local environment issues

Rarely, slow responses trace back to a local issue: a VPN adding latency, a proxy intercepting requests, or a slow DNS resolver. If the status page is clean and your usage is low, run a quick test by disabling any VPN and retrying.

How to diagnose why Claude Code is slow?

Run through this checklist in order. The first match is almost always the cause.

1. Check status.anthropic.com — any active incidents or degraded performance will be listed here. If yes, wait.
2. Run /usage in Claude Code — if you're above 70-80% of your window quota, throttling has likely already started.
3. Check claude.ai/settings/usage — confirms remaining quota across both the 5-hour window and weekly cap.
4. Try a minimal request — send a one-line prompt. If that's also slow, it confirms API-side throttling or a server issue rather than a complex context problem.
5. Disable VPN or proxy — if none of the above explains it, test without any network middleware.

How do usage limits cause slowness in Claude Code?

Claude Code doesn't always stop hard when you hit a limit. On Pro plans especially, Anthropic throttles throughput progressively: responses slow down, then queues increase, then you eventually hit a full block. This staged throttling is intentional and means you can experience significant slowdowns without ever seeing an explicit error. According to Anthropic's support documentation, Pro and Max plans have different usage ceilings, and heavy agentic use drains the window faster than interactive chat.

Symptom	Likely cause	Fix
Responses take 30-60s to start	Quota throttling (70-90% of window used)	Wait for window reset or upgrade to Max
Spinner runs indefinitely	Active API incident or full rate limit hit	Check status page, retry later
Slow on first request of the day	Cold start or server load	Usually self-resolves in minutes
Slows after heavy agentic use	5-hour window nearly exhausted	Check /usage, wait for reset

How to track Claude Code usage before it slows you down?

Usagebar is a macOS menu bar app ($9 one-time) that shows your Claude Code usage in real time: the 5-hour rolling window, the weekly cap, and your context window consumption. It sends notifications at 50%, 75%, and 90% so you know you're approaching a slowdown before it hits. Instead of running /usage manually or getting caught mid-task by throttling, you always have a live read of where you stand.

For more context on how limits work, see Claude Code rate limit exceeded: what it means and does Claude Code usage affect your Pro limits.

Get Usagebar to monitor your Claude Code limits in the menu bar and avoid surprise slowdowns.

What can you do when Claude Code is slow?

Your options depend on the root cause, but here are the practical actions ranked by how often they work:

• Wait for the 5-hour window to reset — if you've burned through your quota, this is the only real fix on the current plan. The window rolls forward from your first request.
• Upgrade from Pro to Max — Max plan users get substantially higher usage limits, which reduces how often throttling kicks in during heavy sessions.
• Reduce context size — large contexts consume more tokens per exchange. Closing and restarting a session clears accumulated context and can restore speed temporarily.
• Avoid peak hours — US East business hours (9am-6pm ET) tend to see the highest API load. Off-peak sessions are consistently faster.
• Check for CLI updates — occasionally a specific Claude Code version has a bug causing unnecessary retries. Run npm update -g @anthropic-ai/claude-code to get the latest.

Is Claude Code faster on Max plan than Pro?

Yes, in practice. The Max plan's higher usage ceiling means you're less likely to hit throttling mid-session. The underlying model and API speed are the same, but Pro users drain their 5-hour window faster on agentic tasks, which is what produces the slowness. If you regularly run multi-file edits or long agentic loops, you'll notice meaningfully fewer slowdowns on Max. See the breakdown in Claude Pro usage limits and reset time.

Sources

• https://status.anthropic.com — Anthropic API status page
• https://support.claude.com/en/articles/11145838-using-claude-code-with-your-pro-or-max-plan — Anthropic: Using Claude Code with Pro or Max
• https://claude.ai/settings/usage — Claude usage dashboard