How Much Does Claude Opus 4.8 Cost? A Complete Guide to API and Subscription Pricing

How API Billing Works

You are charged for the tokens you send to the model (input) plus the tokens the model generates in response (output) . This is a pure pay-as-you-go model — no fixed monthly plans for API access . The model supports a 1-million-token context window at standard pricing with no extra surcharge .

Fast Mode: Speed vs. Cost

Fast Mode on Opus 4.8 runs approximately 2.5 times faster than the standard mode . It costs $10 per million input tokens and $50 per million output tokens . Notably, Anthropic says Fast Mode for Opus 4.8 is now three times cheaper than Fast Mode on previous models .

Caching and Cost Savings

Prompt caching can dramatically reduce your bill. Cache hits cost just $0.50 per million tokens , while writing to cache ranges from $6.25 to $10 depending on the storage duration . Overall, caching can save you up to 90% on input tokens . Batch processing adds another layer of savings — up to 50% off .

Subscription Pricing for Claude Users

Claude Opus 4.8 is available to subscribers on the Pro, Max, Team, and Enterprise plans . The exact cost of these subscriptions is not listed in the provided sources — Anthropic sets those pricing tiers separately on its platform.

Key Takeaways

• All prices quoted are per 1 million tokens .
• Official data is consistent across sources: $5 input / $25 output per million tokens .
• Developers can access the model using the ID claude-opus-4-8 through the Claude API, as well as via Amazon Web Services, Google Cloud, and Microsoft Foundry .