Anthropic’s Claude models are available as fully managed, serverless endpoints on Vertex Artificial Intelligence, eliminating the need to provision or manage infrastructure. Teams can call the models directly through the Vertex Artificial Intelligence API, optionally stream responses via server-sent events to reduce perceived latency, and enable 30-day request-response logging to monitor prompts and completions for potential misuse. Pricing supports pay-as-you-go consumption or fixed-fee provisioned throughput, allowing organizations to align cost with predictable workloads or experimentation.
The lineup covers multiple tiers and use cases. Claude Sonnet 4.5 targets real-world agents with strengths in coding, computer use, cybersecurity, financial analysis, business document handling, and research. Claude Opus 4.1 focuses on frontier intelligence for Artificial Intelligence agents, agentic search and analysis across data sources, expert-level coding, sustained virtual collaboration, long context with memory, and high-quality content creation. Claude Haiku 4.5 delivers near-frontier performance at speed and cost suited to free-tier and latency-sensitive experiences, and is positioned for coding sub-agents, financial analysis at scale, research sub-agents, and business tasks. Earlier Opus 4 and Sonnet 4 models emphasize advanced or efficient coding, long-horizon task execution, agentic search and research, Artificial Intelligence assistants, and large-scale content generation with memory and context management.
Newer models extend reasoning and tool-use capabilities. Claude 3.7 Sonnet introduces “extended thinking,” enabling careful, step-by-step reasoning with a choice between standard and extended modes, and is optimized for agentic coding across the software lifecycle, customer-facing agents, computer use, content generation and analysis, and visual data extraction. Claude 3.5 Sonnet v2 advances real-world software engineering and agentic workflows, including interaction with tools that can manipulate a computer desktop environment, while maintaining the same price and speed as Claude 3.5 Sonnet. Claude 3.5 Haiku prioritizes speed and affordability for code completions, interactive chatbots, data extraction and labeling, and real-time content moderation. Anthropic’s Claude 3 Haiku is the fastest vision and text model for near-instant, human-like interactions spanning live customer support, translations, content moderation, cost-saving operations, and image-to-text analysis. Rounding out the set, Claude 3.5 Sonnet outperforms Claude 3 Opus on Anthropic’s evaluations while retaining mid-tier cost, and is targeted at sophisticated coding, complex customer support, data science and analysis, visual processing, and natural writing. Access to each model is provided through Model Garden model cards within Vertex Artificial Intelligence.
