Claude Sonnet 4.5 API price at a glance
As of November 9, 2026, Anthropic lists Claude Sonnet 4.5 at $3 per million input tokens and $15 per million output tokens on its Claude Platform. Prices are in USD. These are metered API rates—not the price of a Claude.ai subscription. The model reference lists a 200K-token context window; use the rate card below only for requests that fit within that listed context. (Claude Sonnet 4.5 model reference; Anthropic pricing)
Anthropic marks Sonnet 4.5 deprecated and gives November 30, 2026 as its retirement date. It recommends Sonnet 5.5 as a replacement. For a new integration, evaluate the replacement rather than building against a model scheduled to retire. (Claude Sonnet 4.5 model reference)
| Claude API charge | Sonnet 4.5 rate listed for the 200K context scope |
|---|---|
| Input tokens | $3 / million tokens (MTok) |
| Output tokens | $15 / MTok |
| 5-minute prompt-cache write | $3.75 / MTok |
| 1-hour prompt-cache write | $6 / MTok |
| Cache reads | $0.30 / MTok |
| Batch API | 50% off input and output token charges |
Long-context qualification: The model reference lists a 200K context window, but the available official pricing information does not establish whether extended-context access remains available for Sonnet 4.5 on the direct Claude API, or the rates and billing rules for requests beyond that scope. Secondary sources report a 1M option and higher rates, but those terms are not confirmed here. Do not use the standard rates above to budget larger-context requests without checking the current terms for your API channel.
Estimate your Sonnet 4.5 API cost
For standard, uncached usage within the listed 200K context window, calculate input and output separately:
cost = (input tokens ÷ 1,000,000 × $3) + (output tokens ÷ 1,000,000 × $15)
Example: If an application uses 10 million input tokens and 2 million output tokens in a month, with each request fitting within the listed context window:
- Input: 10 × $3 = $30
- Output: 2 × $15 = $30
- Estimated total: $60
This estimate uses the listed per-token rates, assumes standard non-batch usage, and excludes cache charges. Monthly token totals alone do not establish that any individual request exceeds the context window.
Anthropic lists a 50% Batch API discount on input and output charges. If both token categories in this example qualify, the estimate would be $15 for input plus $15 for output, or $30 total. (Claude Sonnet 4.5 model reference)
Is Sonnet 4.5 still available for production?
Anthropic describes Sonnet 4.5 as deprecated: it remains functional but is no longer recommended. The listed retirement date is November 30, 2026, and the recommended replacement is Sonnet 5.5. Anthropic lists Sonnet 5.5 at $2 per million input tokens and $10 per million output tokens. (Claude Sonnet 4.5 model reference)
If you already use Sonnet 4.5, plan and test a migration before the retirement date. Anthropic says retired-model requests fail and advises testing replacement models before migrating. Lifecycle dates can differ on partner-operated platforms such as Amazon Bedrock and Google Cloud, so check the terms for your specific provider. (Model deprecations)
Sources
Claude Sonnet 4.5 — Claude Platform Docs