Anthropic's Claude platform now ships four models: Haiku 4.5, Sonnet 5, Opus 5, and Fable 5.1, reached via direct API, Claude.ai subscriptions, AWS Bedrock, and Google Cloud Vertex AI. Sonnet 5 and Opus 5 both carry a 1 million token context window at standard rates, with no long-context surcharge. Full per-model pricing is in the cost FAQ below.
Anthropic's Claude lineup as of September 2026 spans four models: Haiku 4.5 ($1/$5 per million tokens), Sonnet 5 ($2/$10), Opus 5 ($5/$25), and Fable 5.1 ($10/$50). Sonnet 5 and Opus 5 both support a 1 million token context window at standard rates, and Anthropic cut cache-read costs by 75% for long-running agents this month.
Frequently Asked Questions
What is Anthropic and what does it offer?
Anthropic is an AI safety company founded in January 2021 by Dario Amodei, Daniela Amodei, and five colleagues who previously worked at OpenAI. It provides Claude, a family of large language models, via direct API, Claude.ai subscriptions, AWS Bedrock, and Google Cloud Vertex AI. As of June 2026, the company's annualized revenue run rate is $14 billion and its valuation reached $380 billion in a Series G round closed February 2026. The current model lineup is Haiku 4.5 for speed, Sonnet 5 for production workloads, Opus 5 for complex agentic and coding tasks, and Fable 5.1 for the most demanding reasoning and research work.
How much does the Anthropic Claude API cost?
The Claude API uses pay-as-you-go token billing with no minimum commitment. Claude Haiku 4.5 costs $1 per million input tokens and $5 per million output tokens. Claude Sonnet 5 is $2/$10 per million tokens, the rate Anthropic made permanent on September 1, 2026 instead of raising it. Claude Opus 5 is $5/$25 per million tokens, and Claude Fable 5.1 is $10/$50. The Batch API cuts all token prices by 50% for asynchronous workloads, and Anthropic cut cache-read costs by 75% in September 2026. Claude.ai subscriptions start at $0 (free), $20/month for Pro, $100-$200/month for Max, and $25/seat/month for Team plans.
Does Anthropic offer GPU servers for rent?
No, Anthropic does not sell or rent GPU compute to customers. It uses GPU infrastructure internally to serve Claude model inference, including a major May 2026 compute deal with SpaceX for over 220,000 NVIDIA GPUs. Developers access Claude's capabilities exclusively through the API or subscription interfaces. For raw GPU compute alongside Claude model access, AWS and Google Cloud offer accelerated instances through their Bedrock and Vertex AI integrations respectively.
Who is Anthropic Claude best for?
Claude is best suited for backend engineers and product teams building AI-powered features in production, particularly for code generation, long-document analysis, and autonomous agents. Sonnet 5 and Opus 5 both carry a 1 million token context window at standard rates, making them the strongest choice for legal, financial, or research applications that process full documents in a single request. Claude is not ideal for teams with strict air-gapped or on-premises requirements, since all inference runs on Anthropic or cloud-partner infrastructure. Haiku 4.5 is the better fit for high-volume, latency-sensitive tasks where Opus 5 or Fable 5.1 would be overkill.
Does Anthropic comply with HIPAA and SOC 2?
Yes. Anthropic achieved SOC 2 Type II and ISO 27001 certifications in January 2026, and also holds ISO 42001 certification as an AI management system standard. HIPAA Business Associate Agreements are available on Enterprise plans through sales engagement; any BAA signed after December 2, 2025 covers both the API and Enterprise in one agreement. HIPAA-eligible use excludes certain features: Batch API, Files API, Skills API, extended thinking, Web Fetch, and Computer Use are out of scope under HIPAA. For EU data residency, organizations must route through AWS Bedrock EU or Google Cloud Vertex AI EU endpoints.