Claude Sonnet 5.5 API: Price, Specs, Context and Availability

Claude Sonnet 5.5 specifications

Item	Claude Sonnet 5.5
Release date	September 28, 2026
Official API model ID	claude-sonnet-5-5; APIMaster route ID: claude-sonnet-5.5
Context window	1,000,000 tokens
Maximum output	128,000 tokens in the OpenRouter provider listing
Input / output	Text, image and document input; text output
Standard price	$2 input / $10 output per 1M tokens
Prompt cache	$0.20 reads / $2.50 writes per 1M tokens
Thinking control	Selectable effort; Medium default in Claude apps and Claude Code, High default on Claude Platform
Availability	Claude.ai, Claude Platform, AWS, Google Cloud and Microsoft Azure

The 1M context window is large enough to hold substantial repositories, tool histories and document sets, but context capacity is not the same as reliable recall. Test retrieval quality at the positions and prompt lengths your workflow actually uses. The 128K output ceiling is useful for long reports and code generation, but a high ceiling should not become a default output budget.

What changed from Sonnet 5?

Anthropic calls Sonnet 5.5 a clear upgrade rather than a new price tier. Its headline claims are 30%+ faster generation and up to 30% lower cost per task for most work. The token rate itself did not fall; the saving comes from completing tasks with fewer tokens and fewer steps.

That distinction matters when budgeting. A $2 / $10 model can cost less in practice if it avoids retries, unnecessary tool calls or long reasoning traces. It can also cost more if a workflow always selects the highest effort level and consumes the full output budget. Compare accepted tasks, wall time and total billed tokens, not only list price.

Thinking and effort controls

Sonnet 5.5 exposes effort as an operational control. Lower effort answers faster and uses fewer tokens for routine work; higher effort spends more time checking difficult work. Anthropic sets Medium as the default in Claude apps and Claude Code, and High on the Claude Platform.

There is also a migration detail for integrations that previously disabled thinking. Anthropic says those workflows must move to the new between_tools setting, which keeps up-front thinking off while preserving thinking between tool calls. Do not assume an old Sonnet 5 thinking configuration will behave identically after changing only the model ID.

Coding, agents and knowledge work

Coding: Anthropic positions Sonnet 5.5 for everyday feature work, bug fixes, reviews and repository navigation. Its practical appeal is not just generation speed: smaller, reviewable edits and fewer failed tool calls can reduce the time from prompt to accepted change.

Agents: The model targets well-defined investigation, review and drafting tasks that need reliable multi-step execution. Anthropic describes it as its strongest Sonnet for long-horizon tasks, while Opus 5.5 remains the higher tier for the most complex work.

Knowledge and design work: The 1M context window supports large document sets, while the release emphasizes clearer writing, document and slide work, structured diagrams and polished interfaces. Those are provider claims, so evaluate them against your own templates and acceptance criteria.

Prompt caching and real task cost

Sonnet 5.5 cache reads cost $0.20 / 1M tokens, one tenth of uncached input, while cache writes cost $2.50 / 1M. A repeated system prompt or stable repository prefix can benefit from caching, but only when requests preserve the cacheable prefix and actually hit the cache.

For a fair comparison, record uncached input, cache writes, cache reads, output tokens, retries, latency and human review. Anthropic also advertises up to 50% savings with batch processing. US-only inference is priced at 1.1x for input and output, so deployment geography can change the final bill.

Sonnet 5.5 vs GPT-6 Sol

Both models have the same $2 / $10 headline list price and million-token-scale context, but their product controls differ. Sonnet 5.5 is the Claude-native option with effort control, Claude Code integration and Anthropic's coding and agent behavior. GPT-6 Sol exposes none, low, medium, high, xhigh and max reasoning levels and is already available through 12 APIMaster routes.

Use Sonnet 5.5 when Claude-native behavior and ecosystem compatibility matter. Use GPT-6 Sol when you need an APIMaster route now and want explicit reasoning budgets. The full Sonnet 5.5 vs GPT-6 Sol comparison covers context, cache economics, coding and agent tradeoffs.

How to prepare an integration

1. Confirm the exact model ID is claude-sonnet-5-5 in the provider's current catalog.
2. Check the route, price, region, context limit and output limit before sending production traffic.
3. Migrate thinking-off workflows to the documented between_tools behavior.
4. Run a controlled test set for coding, tools, vision or documents instead of relying on the response's model string.
5. Read how to verify a real Claude Sonnet 5.5 API route before using a third-party relay.

APIMaster currently shows one active Sonnet 5.5 route. Use the model card and send requests to https://apimaster.ai/v1 with the APIMaster route ID claude-sonnet-5.5. Keep the official hyphenated ID claude-sonnet-5-5 when calling Anthropic directly.

FAQ

Is Claude Sonnet 5.5 released?

Yes. Anthropic released Claude Sonnet 5.5 on September 28, 2026. It is available on Claude.ai, Claude Platform and major cloud platforms.

What is the Claude Sonnet 5.5 API model ID?

Anthropic's official model ID is claude-sonnet-5-5. APIMaster's live compatible route is published as claude-sonnet-5.5.

How much does Claude Sonnet 5.5 cost?

Anthropic's standard price is $2 per 1M input tokens and $10 per 1M output tokens. Cache reads are $0.20 and cache writes are $2.50 per 1M tokens. US-only inference carries a 1.1x input/output multiplier.

What are the context and maximum output limits?

Anthropic advertises a 1M-token context window. OpenRouter's provider listing reports a 128K maximum output. Confirm the limits on the exact provider and route you plan to use.

Can I call Sonnet 5.5 on APIMaster now?

Yes. APIMaster listed eight active claude-sonnet-5.5 routes on September 29, 2026, at $0.6691 / $3.34548 per 1M input/output tokens. Use the exact route ID shown on the live card.

Is Sonnet 5.5 cheaper than Sonnet 5?

The token rates are the same. Anthropic says Sonnet 5.5 costs up to 30% less per typical task because it uses fewer tokens and completes work in fewer steps.

Sources and update date

Checked September 29, 2026.

• Anthropic: Introducing Claude Sonnet 5.5 — release date, speed, task-cost claims, prices, defaults, availability and model ID.
• Anthropic: Claude Sonnet — 1M context, use cases, availability and API pricing.
• Anthropic Sonnet 5.5 migration guide — thinking-off migration and between_tools behavior.
• OpenRouter model catalog — 1M context, 128K maximum output and input modalities.
• APIMaster marketplace data checked September 29, 2026 — eight active claude-sonnet-5.5 routes at $0.6691 / $3.34548, plus the current Sonnet 5, GPT-6 Sol and Opus 5.5 prices. The live model cards take precedence.

Start with a live route

Create an APIMaster account, create a key in the console, and open the live Sonnet 5.5 model card. Send requests to https://apimaster.ai/v1 with model ID claude-sonnet-5.5; use the official hyphenated ID only when calling Anthropic directly.