Anthropic’s Claude Opus 5.5 equals Fable at 60% cost

Anthropic introduced Claude Opus 5.5 and reported that the model matches the performance of a comparable Fable offering while costing about 60% per unit. The company said Opus 5.5 will be available through its existing API and enterprise channels, with volume discounts and contract pricing for large customers.

Anthropic published comparative results showing parity on internal benchmarks and in customer pilots. The firm cited tests that covered instruction following, multi-turn dialogue, long-form summarization and code completion. In Anthropic’s examples, pricing expressed as cost per API call or per token of usable output equates to roughly 60% of the figures shown for Fable.

The company reported it validated Opus 5.5 against internal benchmarks and early customer deployments prior to wider release, and provided example workloads and cost comparisons to illustrate potential savings. Anthropic noted that the amount saved will vary by workload type, token volume and latency requirements, and that customers with heavy inference loads may see the largest absolute reductions in spend.

Opus 5.5 integrates with the same developer tools and content-safety filters used by prior Claude releases. Customers can migrate existing workloads using the same API calls and deployment patterns the Claude family already supports.

Anthropic’s materials focused on direct performance and price comparisons rather than architectural differences. The Claude family is designed for conversational AI and task-oriented language tasks and competes with multiple offerings from other startups and established cloud providers.