Claude API

The Claude API is Anthropic's developer platform for accessing the Claude family of large language models programmatically. It provides access to multiple model tiers—Opus, Sonnet, and Haiku—each tuned for different balances of intelligence, speed, and cost. The API is available directly through Anthropic's platform and also through cloud partners including Amazon Bedrock, Google Cloud's Vertex AI, and Microsoft Foundry.

What It Is

The Claude API gives developers and enterprises programmatic access to Claude models for building AI-powered applications, agents, and workflows. Rather than a standalone product, it is the underlying infrastructure layer that powers everything from custom chatbots and coding assistants to large-scale agentic pipelines. Anthropic positions the API as suitable for both early-stage piloting and production-scale deployment.

Model Lineup and Architecture

The API exposes three current model families, each with distinct performance profiles:

• Opus — Anthropic's most intelligent model, optimized for complex agentic tasks and coding
• Sonnet — Designed for an optimal balance of intelligence, cost, and speed
• Haiku — The fastest and most cost-efficient option for high-throughput or latency-sensitive workloads

All models support prompt caching (with 5-minute and extended 1-hour TTL options), which reduces costs on repeated context. A batch processing mode is available for asynchronous workloads, and Anthropic states it saves 50% compared to standard synchronous calls.

Platform Features Beyond Raw Inference

Beyond token-in/token-out inference, the Claude Platform includes several composable capabilities:

• Managed Agents — A suite of APIs for building and deploying agents at scale, billed per session-hour of active runtime in addition to standard token rates
• Web search — Gives Claude access to live web results, billed per search
• Code execution — Runs Python in a sandboxed environment; Anthropic provides a daily free allocation per organization, with additional hours billed separately
• Service tiers — Priority, Standard, and Batch tiers let teams balance availability, performance, and cost predictability

US-only inference is available at a modest premium for workloads with data residency requirements.

Deployment and Integration Paths

Developers access the API through Anthropic's own console at platform.claude.com, or through major cloud marketplaces. The multi-cloud availability via Amazon Bedrock, Google Cloud Vertex AI, and Microsoft Foundry means teams can use Claude within existing cloud procurement and compliance frameworks. Regional compliance options are documented separately for organizations with geographic data requirements.

Update: Claude 4 Model Generation

The pricing page lists Claude Opus 4, Sonnet 4, and Haiku 4 as the current generation, alongside newer point releases such as Opus 4.1, 4.5, 4.6, and 4.7 and Sonnet 4.5 and 4.6. The presence of multiple versioned releases signals active model iteration. A "Mythos preview" model is also listed in the product navigation, suggesting Anthropic is developing a next-generation model family beyond the current Claude 4 lineup. Legacy models remain accessible with their own pricing, giving teams time to migrate at their own pace.