Verified against Anthropic documentation & release specifications on · Updated
Claude Sonnet 5.5 usage limits: how quota, effort, and resets work
Claude Sonnet 5.5 does not have an isolated usage quota. On paid plans (Pro, Max, Team), Sonnet 5.5 usage draws against your plan's rolling 5-hour session allowance and standard weekly plan limit. Anthropic documents a weekly usage limit that applies across all models. Check Settings › Usage for your account's current weekly usage, reset time, and any additional model- or feature-specific limits shown for your account. On the Claude Free tier, Sonnet 5.5 operates under a usage limit that resets every five hours, with no published fixed message allowance.
When you reach your 5-hour session limit, prompts pause until the stated reset time. When you reach your weekly plan limit, switching between Sonnet 5.5 and other models does not bypass the limit because the weekly limit applies across all models. Sonnet 5.5 supports adaptive thinking and defaults to medium effort in Claude web, desktop, and Claude Code (high on the Claude API). Higher effort levels consume more tokens and deplete your allowance faster. You can check remaining quota in Settings › Usage or via /usage in Claude Code, and continue past limits using optional Usage Credits.
Sonnet 5.5 shares your subscription allowance rather than carrying a separate meter. The weekly limit applies across all models, and Settings › Usage shows your account's active weekly progress.
Free tier accounts can use Sonnet 5.5 on claude.ai and mobile apps under limits that reset every 5 hours. Claude Code and higher session allowances require a paid plan.
Sonnet 5.5 supports adaptive thinking across five levels. Claude.ai and Claude Code default to medium effort, while the Claude API defaults to high. Lower effort stretches usage.
On paid plans and Claude Code, a 1,000,000-token context window lets you load large codebases or documents, but longer conversations can consume usage faster because more context must be processed and context management carries a usage cost.
How Claude Sonnet 5.5 usage limits work: session vs. weekly
When you use Sonnet 5.5 on a paid Claude plan, your requests draw from two distinct limit mechanisms with separate time windows:
The rolling 5-hour session limit
The 5-hour session limit regulates short-term message volume across Claude on the web, desktop apps, and Claude Code. Anthropic documents that session allowances reset every five hours. When you reach this limit, Claude displays the exact reset timestamp (such as 5-hour limit reached - resets 4:20 PM).
Read our dedicated 5-hour limit guide for session window mechanics and pacing advice.
The weekly plan limit (all models)
On Pro, Max, and Team plans, weekly limits reset once every seven days at a fixed day and time assigned to your account. Anthropic documents a weekly usage limit that applies across all models. Check Settings › Usage for your account's current weekly usage, reset time, and any additional model- or feature-specific limits shown for your account (Settings › Usage can also show Fable-specific weekly usage when Fable is included in the plan).
For complete weekly schedules, see our weekly limit guide.
Sonnet 5.5 limit comparison by subscription plan
| Plan | Price | 5-Hour Session Quota | Weekly Limit Tracking | Over-Limit Continuation |
|---|---|---|---|---|
Claude FreeFree Tier | $0 | Usage limit that resets every 5 hours | No published weekly limit | Prompts pause until 5h reset; upgrade required for Claude Code |
Claude ProIndividual | $20/mo | More per-session usage than Free | Weekly limit applies across all models | Pauses until reset; optional Usage Credits ($2/M in, $10/M out) |
Claude Max 5×Power Tier | $100/mo | 5× Pro 5-hour session capacity | Weekly limit applies across all models | Pauses until reset; optional Usage Credits |
Claude Max 20×Maximum Tier | $200/mo | 20× Pro 5-hour session capacity | Weekly limit applies across all models | Pauses until reset; optional Usage Credits |
Claude Team (Standard)Team Standard | $25/seat/mo | 1.25× Pro session capacity (per member) | Per-member weekly limit applies across all models | Pauses until reset; org-managed Usage Credits |
Claude Team (Premium)Team Premium | $125/seat/mo | 6.25× Pro session capacity (per member) | Per-member weekly limit applies across all models | Pauses until reset; org-managed Usage Credits |
Enterprise (Usage-Based)Enterprise | Custom | No plan or seat-level session cap | No weekly allowance cap | Billed on actual consumption at standard API rates, subject to admin spend controls |
Enterprise (Seat-Based Legacy)Legacy | Contract | Seat-level limits apply per contract | Per-seat weekly limits until migrated | Pauses until reset; usage credits possible where enabled |
For full cross-plan capacity comparisons across all models and tiers, see our Claude usage limits reference table.
What changed with the Sonnet 5.5 release
Released on September 28, 2026 as the second model in the Claude 5.5 family, Sonnet 5.5 introduced architectural and operational updates that directly affect task completion and quota usage:
Anthropic documents that Sonnet 5.5 generates responses over 30% faster than Sonnet 5 and accomplishes standard developer tasks with fewer total tokens and tool calls, reducing effective task completion costs by up to 30%.
Sonnet 5.5 supports adaptive thinking. In Claude.ai and Claude Code, the model default is medium effort, while the Claude API defaults to high. On the API, up-front thinking can be bypassed using between_tools at low, medium, or high effort.
Sonnet 5.5 scored 70.6% on Terminal-Bench 4.0 (compared to Sonnet 5's 10.3% and Opus 5.5's 66.4%). This performance allows developers to rely on Sonnet 5.5 for agentic coding and multi-file debugging without defaulting to Opus.
Anthropic confirms a 1M token context window for Sonnet 5.5 on paid plans in chat and Claude Code. The Claude Platform model specification also supports a 1M context window and up to 128,000 output tokens, alongside Opus-tier cyber safeguards.
Reasoning effort levels: how adaptive thinking affects your quota
In Claude apps and Claude Code, the effort setting is your primary dial for controlling how fast you consume your 5-hour and weekly allowances. Anthropic says higher effort uses more tokens and reaches usage limits faster, while lower effort preserves quota.
Sonnet 5.5 vs. Opus 5.5: usage and efficiency comparison
Deciding when to use Sonnet 5.5 versus Opus 5.5 is often a quota-management decision. Here is how both models compare across documented dimensions:
| Dimension | Claude Sonnet 5.5 | Claude Opus 5.5 |
|---|---|---|
| Plan Availability | Claude Free (web/mobile), Pro, Max, Team, Enterprise | Paid plans only (Pro, Max, Team, Enterprise) and API |
| Weekly Limit Scope | Applies across all models; tracked in Settings › Usage | Applies across all models; tracked in Settings › Usage |
| App & CLI Effort Default | medium | medium |
| Standard API Pricing | $2.00/M in · $10.00/M out | $4.00/M in · $20.00/M out |
| Prompt Cache Read Price | $0.20/M tokens | $0.20/M tokens |
| Context Window (Platform Spec) | 1M tokens on paid plans & API (128k max output on Platform) | 1M tokens on paid plans & API (128k max output on Platform) |
| Terminal-Bench 4.0 Score | 70.6% | 66.4% |
Important distinction on quota multipliers: At API token rates, Sonnet 5.5's input and output prices are half of Opus 5.5's ($2/$10 vs $4/$20), while cache reads cost the same $0.20 per million tokens. However, Anthropic does not publish a universal 2× subscription quota multiplier between Sonnet and Opus.
Do not assume your monthly subscription provides twice as many messages simply because token rates are 50% lower. While Sonnet 5.5 finishes standard tasks with fewer tokens and tool calls, subscription rate limiters operate dynamically based on session complexity and server load.
What we measured on one task: in our Claude Code cost study, Sonnet 5.5 High built the same Next.js feature for a median of $0.0925 in API-equivalent cost, against $0.4456 for Opus 5.5 High, about 0.21 times. Lower prices explain only part of that: in the cache-controlled stage Sonnet also made 4 or 5 model requests where Opus made 12 to 20, and it never ran the app in a browser to check its work. Both passed the same functional assessment, but they handled one edge case differently, so read it as a cost comparison for one task, not a quality ranking or a quota multiplier.
For dedicated coverage of Opus 5.5 limits, see our complete Claude Opus 5.5 usage limits guide.
The 1M context window: practical usage realities
Anthropic supports a 1,000,000-token context window for Sonnet 5.5 on paid Claude plans and in Claude Code. While this allows loading large documentation sets, multiple repository folders, or extensive datasets into a conversation, having a larger context window does not grant you a larger subscription quota:
Context capacity is not quota: The context window measures how much text the model can process in a single interaction. Subscription limits measure your cumulative consumption over rolling 5-hour sessions and 7-day weekly periods.
Why long conversations consume usage faster: Long conversations can consume usage faster because more context may need to be processed and context management itself has a usage cost. Prompt caching can reduce the cost of repeated context, so context-window size cannot be converted directly into subscription-quota consumption with a simple tokens-per-turn formula. Anthropic does not publish a fixed subscription token accounting formula.
In Claude chat, automatic context management summarizes earlier conversation content for users with code execution enabled when threads near capacity. Even with caching and summarization, Anthropic confirms that longer conversations and active context management draw more heavily against your overall usage limits.
To avoid premature lockouts, treat the 1M window as a tool for focused investigations. When moving to an unrelated task or feature, start a fresh conversation or clear your session context.
Sonnet 5.5 in Claude Code vs. web chat
Subscription logins share the same 5-hour and weekly quotas between web chat and Claude Code. However, agentic CLI coding sessions tend to consume quota much faster than browser conversations:
Web Chat Behavior
- •Context consists primarily of user prompts, uploaded documents, and model answers.
- •Each interaction is typically a single turn without autonomous tool execution loops.
- •Quota drains predictably based on message length and file attachments.
Claude Code Terminal Behavior
- •Automatically reads repository files, runs grep commands, and inspects directory trees.
- •Executes multi-step agent loops where tool outputs are repeatedly added to context.
- •Spawning parallel subagents multiplies token consumption across concurrent model calls.
Best practices to preserve Sonnet 5.5 quota in Claude Code:
- 1.Check remaining allowance frequently: Run
/usageto monitor your active 5-hour session bar, 7-day weekly bar, and exact reset time. Learn more in our Claude Code usage guide. - 2.Compact context between milestones: Run
/compactto summarize past tool output once a task milestone is reached. - 3.Clear context between features: Use
/clearto reset conversational memory before tackling an unrelated bug. - 4.Keep CLAUDE.md concise: Keep project instruction files lean to prevent hundreds of lines of static instructions from bloating every prompt turn.
- 5.Use an API key for headless automation: If running automated scripts or bulk tests, authenticate Claude Code with an
ANTHROPIC_API_KEYto bill tokens directly without exhausting your interactive subscription limits.
Where to check your Sonnet 5.5 usage and reset times
You can verify remaining quota and scheduled reset times across three primary surfaces:
1. Claude Web & Desktop Apps (Settings › Usage)
Click your avatar in the lower-left corner and open Settings › Usage. Paid accounts display progress toward the rolling 5-hour session allowance and weekly limit reset information across all models, along with scheduled reset times and any feature-specific limits shown for your account (such as Fable-specific weekly usage when Fable is included in the plan).
2. Claude Code CLI (/usage command)
Run /usage in any active terminal session. For subscription accounts, Claude Code reports 5-hour session progress, 7-day weekly quota status, and upcoming reset timestamps.
3. Terminal Statusline & Menu Bar
To keep usage visible while coding, configure a Claude Code statusline or install the free, open-source local Claude usage tracker to monitor your allowances directly from your macOS menu bar or CLI prompt.
What happens when you hit a Sonnet 5.5 limit
Your options when reaching a limit depend on which boundary was triggered:
Prompts pause until the rolling 5-hour reset timestamp displayed in the interface. When that time arrives, your session allowance restores automatically.
Anthropic documents that weekly plan limits apply across all models. Check Settings › Usage for your account's current weekly usage, reset time, and any additional model- or feature-specific limits shown for your account.
With optional Usage Credits enabled in Settings › Usage, Claude continues processing Sonnet 5.5 requests without waiting for the reset window. Continuation is billed based on actual token consumption at standard Sonnet 5.5 API rates ($2.00 per million input tokens, $10.00 per million output tokens, and $0.20 per million cache read tokens).
For setup instructions, spend caps, and bundle discounts, read our Claude usage credits guide. If currently locked out, check our limit reached checklist.
Frequently asked questions about Sonnet 5.5 usage limits
Does Claude Sonnet 5.5 have its own separate usage limit?
No. Claude Sonnet 5.5 does not have an isolated subscription quota. On paid plans (Pro, Max, and Team), Sonnet 5.5 usage draws against your plan's rolling 5-hour session allowance and standard weekly plan limit. Anthropic documents a weekly usage limit that applies across all models. Check Settings › Usage for your account's current weekly usage, reset time, and any additional model- or feature-specific limits shown for your account. On the Claude Free tier, Sonnet 5.5 operates under a usage limit that resets every five hours, with no published fixed message or token allowance.
Is Claude Sonnet 5.5 available on the free Claude plan?
Yes. Anthropic makes Claude Sonnet 5.5 available on the Claude Free tier in web and mobile apps. Free accounts are governed by usage limits that reset every five hours, without a published weekly allowance. Anthropic does not publish a fixed message or token allowance for the Free tier. Accessing Sonnet 5.5 in Claude Code requires an active paid subscription (Claude Pro, Max, Team, or Enterprise) or pay-as-you-go Anthropic API credentials.
How many messages or prompts do you get with Sonnet 5.5?
Anthropic does not publish fixed message quotas or universal token ceilings for Sonnet 5.5. Your practical capacity depends on conversation length, file attachments, reasoning effort level, and tool calls. Longer conversations and deep agentic sessions can consume your allowance faster because more context needs to be processed and context management carries a usage cost, though prompt caching can reduce the impact of repeated context.
Does switching between Sonnet 5.5 and Opus 5.5 bypass an exhausted weekly limit?
No. Anthropic's Pro, Max, and Team plan documentation explicitly states that the weekly usage limit applies across all models. Switching between Sonnet 5.5 and Opus 5.5 does not bypass an exhausted weekly limit. Check Settings › Usage for your account's current weekly usage, reset time, and any additional model- or feature-specific limits shown for your account.
What is the default effort level for Sonnet 5.5?
In Claude web, desktop apps, and Claude Code, Sonnet 5.5 defaults to 'medium' effort. On the Claude API, the default effort is 'high'. API users can also set thinking to between_tools at low, medium, or high effort to bypass up-front thinking. You can adjust effort across five levels (low, medium, high, xhigh, max); lower effort generates shorter reasoning traces and stretches your usage allowance further.
Does Claude Code share the same Sonnet 5.5 limit as web chat?
Yes. When you log into Claude Code with your Claude Pro, Max, or Team subscription, terminal commands and browser chats draw from the exact same 5-hour session and weekly plan limits. If you authenticate Claude Code with an API key instead of your subscription login, usage is billed per token through the Anthropic Console and does not touch your subscription quota.