Anthropic lists Claude Opus 4.5 at $5 per million input tokens and $25 per million output tokens through the Claude API. The model ID is claude-opus-4-5-20251101. These are Anthropic’s direct Claude API rates; other providers and routes may have different prices or billing terms. (Anthropic pricing)
Claude Opus 4.5 API pricing by usage type
Anthropic lists these USD rates per million tokens (MTok) for Claude Opus 4.5: (Anthropic pricing)
| Claude API usage | Price per 1M tokens | Notes |
|---|---|---|
| Standard input | $5 | Direct Claude API rate. |
| Standard output | $25 | Direct Claude API rate. |
| Prompt-cache write, 5-minute duration | $6.25 | Priced separately from standard input. |
| Prompt-cache write, 1-hour duration | $10 | Priced separately from standard input. |
| Prompt-cache read | $0.50 | Applies to cache hits and refreshes. |
| Batch API input and output | 50% discount | Anthropic lists this discount for Batch API input and output. |
These are Claude API rates, not a universal price for every way to run Opus 4.5. Anthropic’s model reference identifies the model as Active (legacy) and says it remains available. (Claude Opus 4.5 model reference)
Amazon Bedrock is billed separately. AWS says Claude Opus 4.5 on Bedrock is offered and billed through AWS Marketplace, with charges appearing on the AWS bill and Cost Explorer under the model provider. Check AWS pricing for the specific offer and deployment rather than assuming Anthropic’s direct API rates apply. (AWS Bedrock model card)
Estimate your Claude Opus 4.5 API spend
For standard, uncached Claude API usage, estimate input and output separately:
Estimated token cost = (input tokens ÷ 1,000,000 × $5) + (output tokens ÷ 1,000,000 × $25)
Example: If an application uses 10 million standard input tokens and generates 2 million output tokens in a month:
- Input: 10 × $5 = $50
- Output: 2 × $25 = $50
- Estimated token spend: $100
This example uses the listed standard Claude API rates and assumes no prompt-cache or Batch API pricing. It estimates token charges, not a guaranteed invoice total. For cache-enabled requests, calculate cache writes at the rate for the selected cache duration and cache reads at their listed rate. For eligible Batch API requests, account for the listed input/output discount. (Anthropic pricing)
Production API checks before you budget
- Confirm the route and model ID. Anthropic lists
claude-opus-4-5-20251101for the Claude API. Bedrock uses a separate route and billing arrangement, so verify its offer and regional terms independently. (Anthropic model reference; AWS Bedrock model card) - Check lifecycle plans before a new commitment. Anthropic’s model reference identifies Opus 4.5 as Active (legacy), says it remains available, and recommends considering migration to Opus 5.5. Check lifecycle guidance for your intended route; Amazon Bedrock sets its own retirement schedule. (Anthropic model reference; Anthropic model deprecations)
- Validate Bedrock limits and cache behavior for your workload. AWS lists a 200K-token context window and explicit prompt caching for Opus 4.5 on Bedrock. Its model card specifies a 4,096-token minimum per cache checkpoint, up to four checkpoints per request, and 5-minute or 1-hour TTLs. AWS explains that the minimum applies cumulatively to the prompt prefix before each checkpoint. (AWS Bedrock model card; AWS prompt-caching documentation)
Bottom line
For the Anthropic Claude API, the listed standard rates for Claude Opus 4.5 are $5 per million input tokens and $25 per million output tokens. At those rates, 10 million input tokens plus 2 million output tokens comes to $100 before other billing adjustments. Cache and Batch API rates are separate, and Amazon Bedrock has its own billing route. Anthropic labels Opus 4.5 a legacy model, so check the lifecycle guidance for your chosen route before making a longer-term production commitment. (Anthropic pricing; Anthropic model reference)
Sources
Claude Opus 4.5 — Claude Platform Docs
Model deprecations — Claude Platform Docs
Claude Opus 4.5 — Amazon Bedrock