Claude Sonnet 5.5 Explained: Access, Pricing and New Usage Limits

What is Claude Sonnet 5.5?

Claude Sonnet 5.5 is the second model in Anthropic's Claude 5.5 family, following Claude Opus 5.5.

Anthropic positions Sonnet 5.5 as the faster and lower-cost complement to Opus 5.5. The model is intended especially for work where the task is reasonably well defined and the user cares about iteration speed, responsiveness and cost.

Anthropic highlights everyday coding, bug fixing, application development, document creation, presentation slides, spreadsheets, design work and other professional workflows as important Sonnet 5.5 use cases.

That does not mean Sonnet 5.5 is simply a lightweight model. Anthropic's published evaluations show large improvements over Sonnet 5 in coding and professional knowledge work, while the company says Opus 5.5 remains better suited to the most difficult, ambiguous work requiring sustained judgment.

Official source: Anthropic's Claude Sonnet 5.5 announcement .

What changed from Claude Sonnet 5?

Anthropic describes Sonnet 5.5 as a clear upgrade over Sonnet 5 rather than a simple version-number refresh.

The company says Sonnet 5.5 generates output more than 30% faster than Sonnet 5 and typically requires fewer tokens to complete a task.

That efficiency matters in two different ways. Interactive users spend less time waiting during repeated edits, while developers running the model through an API can reduce the amount of inference required to finish many workloads.

The list price therefore did not need to fall for Sonnet 5.5 to become cheaper on many tasks. If the new model finishes the same job using fewer output and reasoning tokens, the total cost of that job can fall even when each token costs the same amount.

How to access Claude Sonnet 5.5

Anthropic says Sonnet 5.5 is available across its current platform ecosystem. The precise models visible to an individual user can still depend on account, plan, workspace policy and rollout state, so the model selector in your own Claude account remains the most reliable check for app access.

Claude apps

In Claude's apps, open a new conversation and inspect the model selector. If Sonnet 5.5 is enabled for your account, select it before beginning the task.

Anthropic says the default effort for Sonnet 5.5 in Claude Code and its apps is Medium.

Claude Platform

Developers using Anthropic's direct API can call the model with:

Cloud platforms

Anthropic also lists Sonnet 5.5 for Amazon Bedrock, Google Cloud, Microsoft Foundry and Claude Platform on AWS. Provider-specific model identifiers and enablement procedures can differ, so developers should use the identifier documented for the platform they are actually calling.

Model availability: Claude Sonnet 5.5 model overview .

Claude Sonnet 5.5 API specifications

The 1M-token context window describes how much material the model can process within an applicable request or product context. It should not be confused with a subscription usage allowance.

A context window limits the amount of information available within a task. A usage limit controls how much Claude activity an account can consume over a period of time. They solve different problems.

Claude Sonnet 5.5 pricing explained

Claude subscription pricing and Claude API pricing are separate systems.

A Claude subscription gives a user access to the Claude application and an included amount of usage according to that plan. Direct Claude API usage is billed according to tokens and the developer organization's API billing arrangement.

Those prices apply to direct API token consumption. They are not the monthly subscription price for Claude Pro, Max or an organization plan.

Pricing source: Claude Sonnet 5.5 Platform documentation .

Why Anthropic says Sonnet 5.5 can cost up to 30% less

The phrase “30% less” can be confusing because Sonnet 5.5 retains the same headline $2 input and $10 output pricing tier as Sonnet 5.

Anthropic's claim is based on cost per completed task, not only price per token.

According to Anthropic, Sonnet 5.5 often completes work using fewer tokens and fewer unnecessary intermediate steps. If a task that previously consumed 100 units of paid processing can now be completed with substantially less processing, the final task cost drops even if the underlying token rate has not changed.

Anthropic also says Sonnet 5.5 produces output more than 30% faster than Sonnet 5. Faster generation improves latency, while lower token consumption influences cost. They are related improvements but should not be treated as the same measurement.

What are the new Claude 5.5 usage limits?

Anthropic announced broader subscription usage changes when the Claude 5.5 family began rolling out with Opus 5.5 on September 22, 2026.

The company said it was increasing five-hour usage limits on Pro, Max, Team and seat-based Enterprise plans.

That announcement applies to the wider Claude subscription allowance. It should not be read as evidence that Sonnet 5.5 has a separate, fixed number of messages that every subscriber receives.

Anthropic's general usage documentation says activity across Claude product surfaces can count against the same underlying subscription allowance. For example, Claude and Claude Code usage can draw from the same plan limits for relevant individual plans.

Sources: Anthropic's Claude 5.5 usage-limit announcement and Understanding usage and length limits .

How the five-hour usage limit works

Claude's paid individual plans use session-based usage allowances that refresh on a five-hour cycle.

Anthropic's Pro documentation explains that the number of messages a user can send is not fixed. Message length, attachments, conversation history, model selection and feature use all affect how much allowance is consumed.

If you reach the applicable five-hour allowance, Claude can display a message showing when the session limit will become available again.

The increased allowance introduced during the 5.5 rollout means eligible plans have more room in that five-hour window, but the window itself still exists.

How to check the limit affecting your account

1. Open Claude using the account associated with your subscription.
2. Open Settings.
3. Open Usage.
4. Review the active usage allowance and the reset time displayed for your account.
5. Check whether the restriction is the five-hour session allowance, a weekly limit or another model-specific allowance before assuming that the model itself is unavailable.

Official guidance: Anthropic Pro plan usage limits .

Weekly limits still matter

A five-hour reset does not necessarily restore every type of Claude allowance.

Anthropic's current Pro documentation says the plan also has a weekly usage limit that applies across models. The weekly allowance resets at a fixed day and time associated with the account.

This explains why a user can occasionally see their five-hour window refresh and still encounter a restriction. The session allowance may be available again while the weekly allowance remains exhausted.

This is also why switching from one Claude model to another does not guarantee that a depleted shared allowance will disappear.

What is the free Claude limit reset?

Anthropic also introduced a separate limit reset for eligible subscribers during the Claude 5.5 rollout.

A limit reset is not the same thing as the normal five-hour automatic refresh. It is a separate entitlement that can restore an applicable usage allowance when the user chooses to activate it.

Anthropic's current support documentation says a reset can apply to either a five-hour session limit or a weekly usage limit, depending on the reset shown to the account.

It also makes several important points:

The reset control currently appears on the web or Claude Desktop where the account is eligible. Anthropic says the reset button itself is not currently available inside Claude Mobile or Claude Code, although restoring a shared allowance can affect those surfaces because the account's usage limit is shared.

Official reference: What is a limit reset? .

Can you continue after reaching your included Claude limit?

Eligible paid subscribers can enable usage credits to continue using Claude after the included subscription allowance has been exhausted.

Usage credits are separate from the subscription's normal included usage and separate from a free limit reset.

Anthropic says subsequent usage can be billed at consumption-based rates once usage credits are enabled and the included allowance has been reached.

1. Open Claude on the web.
2. Open Settings > Usage.
3. Locate the Usage credits section.
4. Enable the feature if you want pay-as-you-go continuation.
5. Configure the payment method, funds and spending controls offered for the account.

Official reference: Manage usage credits for paid Claude plans .

Effort settings affect speed, cost and usage

Sonnet 5.5 supports Anthropic's adaptive thinking system and the effort control used to determine how deeply the model works on a request.

Anthropic says lower effort settings generally produce faster responses and consume fewer tokens, which makes them suitable for routine work. Higher effort allows Claude to reason for longer and check its work more thoroughly.

The default differs by product surface:

This matters when comparing API costs with an interactive Claude session. Two requests using the same model name can consume different amounts of reasoning and output if they run with different effort settings.

It also matters for subscription usage. Anthropic's general usage guidance specifically notes that higher effort can consume more tokens and therefore use an allowance more quickly.

Should you use Sonnet 5.5 or Opus 5.5?

Anthropic presents the two models as complementary.

Sonnet 5.5 is the lower-cost and faster choice for work whose boundaries are already clear. Opus 5.5 is intended for work where understanding the problem, choosing an approach and maintaining judgment over a long workflow are the difficult parts.

BeeVaults covers that question in more detail in Claude Sonnet 5.5 vs Opus 5.5: Which Should You Use? .

Developers should not treat Sonnet 5.5 as only a model-name change

Existing Sonnet 5 integrations can require configuration changes before moving to Sonnet 5.5.

Anthropic's migration documentation lists several behavioral and API changes, especially around thinking configuration, tool choice, content blocks and computer use.

For example, Sonnet 5.5 no longer accepts the older explicit thinking: {"type":"disabled"} pattern. Anthropic uses between_tools when developers need the lowest thinking mode with no up-front thinking.

Sonnet 5.5 also rejects forced tool-choice types that previously used tool or any, and adaptive thinking can cause a thinking block to appear before the first visible text block.

BeeVaults has a dedicated migration guide covering these failures: Claude Sonnet 5.5 API Errors: Fix Thinking and Tool Use Settings .

Official migration reference: Claude Sonnet 5.5 migration guide .

Does the 1M context window mean unlimited conversations?

No.

Context capacity and account usage are separate. A 1M-token context window describes the amount of material Sonnet 5.5 can potentially work with in an applicable model request.

Your Claude plan can still impose session and weekly usage limits. Long conversations, project files, tool activity and large prompts can also consume more of that plan allowance.

Anthropic's usage documentation distinguishes length limits, which constrain how much information fits into a conversation or request, from usage limits, which constrain how much Claude can be used during a period.

A larger context window therefore gives the model more room to understand a large task; it does not turn a subscription into unlimited usage.

Frequently asked questions

What Sonnet 5.5 changes in practice

Claude Sonnet 5.5 keeps Sonnet in the lower-cost $2 input and $10 output API tier while pushing the model substantially closer to Anthropic's higher-end models on coding and professional work.

The practical advantage is not merely benchmark performance. Anthropic says the model completes many tasks with fewer tokens and produces output more than 30% faster than Sonnet 5, which can improve both responsiveness and total task economics.

For Claude subscribers, the wider 5.5 rollout also brought larger five-hour allowances to several paid plans and introduced limit-reset functionality for eligible accounts. Those changes do not eliminate weekly limits, do not create unlimited Claude use and should not be confused with a separate fixed Sonnet 5.5 message quota.

Developers should also read Anthropic's migration guidance before moving a production integration from Sonnet 5 to Sonnet 5.5 because the new model changes several request and response behaviors around thinking and tool use.