What Is the Claude Session Limit? (5-Hour Window, Weekly Cap, Context Explained)

The Claude session limit is a rolling 5-hour usage window that caps how much you can use Claude Code within any given 5-hour period. Once you hit it, Claude pauses your access until the window resets. There is also a separate weekly cap that applies across all plans. Pro users get a lower ceiling; Max 5x and Max 20x users get progressively more. The limit is not a fixed daily message count — it is usage-based and tied to model weight and token consumption.

• 5-hour rolling window resets from when you started using, not at midnight
• Weekly cap runs independently of the 5-hour window — you can hit either first
• Opus 5 and Sonnet 5 consume the limit faster than lighter models

What is the 5-hour session window?

The 5-hour window is a sliding period that starts when you send your first message in a session. Anthropic tracks cumulative usage within that window and throttles access once you exceed the plan's threshold. The window is rolling, not fixed to a clock — so if you start at 2:00 PM, it resets around 7:00 PM regardless of time zone or date. This is documented in Anthropic's Claude Code plan limits article.

When the 5-hour limit is hit, Claude Code displays a message telling you how long until access resumes. You cannot bypass this by starting a new conversation — the window is account-level, not conversation-level.

How does the weekly cap work?

Separate from the 5-hour window, each plan has a weekly usage cap. This resets on a 7-day rolling basis tied to your billing cycle. Heavy users — especially those running long agentic tasks or multi-file refactors — can exhaust the weekly cap before the 5-hour limit becomes relevant. See the weekly limit vs. 5-hour lockout breakdown for a deeper comparison of how each fires first depending on usage patterns.

What are the session limits by plan?

As of July 2026, Anthropic does not publish exact token thresholds for the 5-hour window, but the relative scale across plans is clear from official documentation:

Plan	Relative usage limit	Best for
Claude Pro ($20/mo)	Baseline (1x)	Light to moderate daily use
Claude Max 5x ($100/mo)	~5x Pro	Full-time developer use
Claude Max 20x ($200/mo)	~20x Pro	Power users, agentic workflows

Pro plan users hit the 5-hour limit quickly during heavy refactoring sessions. Max 5x is the recommended tier for developers using Claude Code as their primary coding tool. For a full comparison, see Claude Pro vs. Max limits.

What counts toward the session limit?

Usage is weighted by model and token volume. A single long Opus 5 task burns significantly more of your session budget than an equivalent Sonnet 5 task. The following all count toward the limit:

• Every prompt and response (input + output tokens)
• Context injected into long conversations (files, tool call results)
• Agentic loops where Claude reads, edits, and re-reads files
• Background agent tasks running in parallel

Context accumulation is a major hidden driver. As a conversation grows, the injected context on every message grows with it. This is why multi-turn debugging sessions drain limits faster than expected. The Claude Code context window explained post covers this in detail.

How is the session limit different from the context window limit?

They are separate constraints. The session limit is a rate/usage cap managed at the account level — it measures cumulative consumption over time. The context window limit is a per-conversation constraint: Claude can only hold ~200,000 tokens in a single conversation window before older content gets truncated or the conversation must be restarted. You can hit either independently. A short conversation with heavy tool calls can exhaust your session limit; a long multi-turn thread can exceed the context window without ever hitting the session cap.

How do you check current session usage?

There are three ways to monitor where you stand:

• /usage command: Type /usage inside Claude Code to see a summary of your current window consumption and time until reset.
• claude.ai settings: Go to claude.ai/settings/usage to view usage statistics for your account.
• Usagebar: A $9 one-time macOS menu bar app that shows your 5-hour window, weekly cap, and context usage in real time with notifications at 50%, 75%, and 90% thresholds — so you're never caught off guard mid-session.

The /usage command is fine for a spot check, but it requires you to already be in Claude Code. Usagebar is more useful during active work because it surfaces the data passively without interrupting your flow. More on checking limits: how to check Claude Code usage limits.

What happens when you hit the session limit?

Claude Code displays a rate limit message and stops accepting new prompts for your account until the 5-hour window resets. Ongoing agentic tasks are interrupted. The error typically includes a timestamp for when access will resume. See what to do when Claude Code rate limit is exceeded and how to fix the message limit exceeded error for recovery steps.

How do you avoid hitting the session limit early?

A few practical changes reduce consumption meaningfully:

• Use Sonnet instead of Opus for routine tasks: Opus 5 is heavier; use it only for complex reasoning tasks.
• Compact long conversations: Use /compact to summarize context and reduce token injection on each turn.
• Scope your tasks: Avoid open-ended "refactor everything" prompts. Smaller, targeted requests use less context per turn.
• Monitor proactively: Usagebar's 75% notification gives you a heads-up before you're blocked, so you can wrap up critical work first.

Key takeaways

1. The Claude session limit is a rolling 5-hour window, not a daily or per-conversation count.
2. A separate weekly cap applies on top of the 5-hour limit — you can hit either first.
3. Max 5x ($100/mo) is the practical minimum for full-time Claude Code use.
4. Context accumulation in long conversations silently accelerates limit consumption.
5. Check usage with /usage, claude.ai/settings/usage, or Usagebar for real-time menu bar tracking.

Sources

• Anthropic: Using Claude Code with your Pro or Max plan
• claude.ai usage settings