Anthropic

Claude is also distributed through Anthropic-operated marketplace channels (Claude Platform on AWS and Claude in Microsoft Foundry, both billed in Claude Consumption Units at standard Anthropic per-token rates) and partner-operated clouds (Amazon Bedrock and Google Cloud Vertex AI, which set their own prices and lifecycle dates). Positioning is frontier reasoning and long-horizon agentic coding rather than breadth of modality: no first-party image/audio/video generation, no embeddings model, and no open-weight releases.

A Access and limits

Sign up at platform.claude.com and create a key under Settings > API keys; authenticate with the x-api-key header plus anthropic-version: 2023-06-01, or with an OAuth profile via ant auth login (SDKs pick the profile up automatically). New organizations receive a small amount of free trial credit; there is no ongoing free tier and no free-forever model. Usage tiers gate both rate limits and a monthly spend cap: Start ($500/mo cap), Build ($1,000), Scale ($200,000), and Custom (no cap, negotiated).

Example Scale-tier limits: Opus 5 and Sonnet 5 at 10,000 RPM / 10M input TPM / 2M output TPM each; Fable 5.x shares a combined 4,000 RPM / 4M ITPM bucket. Cache-read tokens do not count toward ITPM on any current model (Haiku 3.5 is the exception). PRICING MODIFIERS, all recorded prices are the base/standard tier: Batch API is 50% off input and output; 5-minute cache writes cost 1.25x base input and 1-hour writes 2x; cache reads cost 0.1x base input, except Claude Fable 5.1 and Claude Mythos 5.1 where reads are 0.025x ($0.25/MTok).

There is NO long-context surcharge - Claude 4.6-generation models and later include the full 1M-token window at standard per-token rates (a 900k-token request costs the same per token as a 9k one). inference_geo: "us" (US-only inference, Claude 4.6+) applies a 1.1x multiplier to every token category; global routing is the default and standard price. Fast mode (research preview, Claude API only, Opus 5 and Opus 4.8 only) is billed at $10/$50 per MTok. Server-tool add-ons: web search $10 per 1,000 searches; web fetch free beyond token costs; code execution free when paired with the 2026-era web search/fetch tools, otherwise 1,550 free container-hours per org per month then $0.05/hour/container; Managed Agents adds $0.08 per session-hour on top of tokens.

Regions: the first-party API is global by default with optional US-only inference; Bedrock and Google Cloud offer global, multi-region and regional endpoints, with regional carrying a 10% premium. An OpenAI-SDK compatibility layer exists at base URL https://api.anthropic.com/v1/ for /chat/completions, but Anthropic explicitly frames it as a testing/evaluation shim, not a production path - prompt caching, strict tool schemas, response_format, audio and logprobs are all ignored or unsupported there. Claude Mythos 5.1 / Mythos 5 / Mythos Preview are invitation-only under Project Glasswing (defensive cybersecurity); contact an Anthropic, AWS or Google Cloud account team.

Claude Fable-tier models require 30-day data retention and are not available to zero-data-retention orgs without express authorization.

C Calling the API

Anthropic accepts requests in the OpenAI Chat Completions format, so any OpenAI-compatible client works by changing the base URL. Example uses claude-haiku-4-5-20251001.

curl https://api.anthropic.com/v1/chat/completions \
  -H "Authorization: Bearer $API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "claude-haiku-4-5-20251001",
    "messages": [{"role": "user", "content": "Hello"}]
  }'

from openai import OpenAI

client = OpenAI(
    base_url="https://api.anthropic.com/v1",
    api_key=os.environ["API_KEY"],
)

response = client.chat.completions.create(
    model="claude-haiku-4-5-20251001",
    messages=[{"role": "user", "content": "Hello"}],
)
print(response.choices[0].message.content)

A compatibility layer is not always the provider's full API — caching, strict tool schemas and structured output are commonly unsupported there. Check the official docs before relying on an advanced feature.

D Official references

• Homepage
• API documentation
• Pricing