As of November 8, 2026, Anthropic’s current model overview lists four Claude API IDs: Opus 5.5 (claude-opus-5-5), Sonnet 5.5 (claude-sonnet-5-5), Haiku 5.5 (claude-haiku-5-5), and Fable 5.1 (claude-fable-5-1). Anthropic recommends starting with Opus 5.5 for most workloads. These are direct Claude API IDs, not identifiers for a separate gateway or cloud-provider route. Check the current model overview or GET /v1/models for the latest inventory and availability details.
Claude API model IDs and how to choose
Anthropic lists all four models in its current Claude API lineup. Its overview reports a 1-million-token context window and up to 128,000 output tokens for each. For availability at request time, check the Models API, which returns currently available models and details such as token limits and capabilities.
| Model | Claude API model ID | Listed status | Anthropic’s stated fit | Input / output price per MTok |
|---|---|---|---|---|
| Claude Opus 5.5 | claude-opus-5-5 |
Listed in current Claude API lineup | Long-running agentic coding and knowledge work | $4 / $20 |
| Claude Sonnet 5.5 | claude-sonnet-5-5 |
Listed in current Claude API lineup | Combination of speed and intelligence | $2 / $10 |
| Claude Haiku 5.5 | claude-haiku-5-5 |
Active (latest) | High-volume, latency-sensitive tasks such as classification, extraction, and routing | $0.10 / $0.50 for prompts up to 100,000 tokens; $0.50 / $2.50 above that threshold |
| Claude Fable 5.1 | claude-fable-5-1 |
Listed in current Claude API lineup | Demanding reasoning and long-horizon agentic work | $10 / $50 |
Anthropic’s overview recommends Opus 5.5 as a starting point for most workloads. It positions Sonnet 5.5 as a speed-and-capability option, Haiku 5.5 for high-volume, latency-sensitive tasks, and Fable 5.1 for demanding reasoning or long-horizon agentic work. These are the provider’s selection descriptions, not a guarantee of performance for a particular application. The overview lists the four IDs under Claude API; check the live Models API for the models currently available to your integration.
Make a basic Messages API request
Create an API key in the Claude Console and store it in an environment variable. Anthropic documents API-key authentication with Authorization: Bearer <key> for direct HTTP requests. Requests also require the anthropic-version header; 2026-10-21 is a documented API version. The Messages API uses a JSON request with a model ID, max_tokens, and a messages array. See Anthropic’s API-key instructions, API overview, and versioning reference.
export ANTHROPIC_API_KEY="your-api-key"
curl https://api.anthropic.com/v1/messages \
-H "Authorization: Bearer $ANTHROPIC_API_KEY" \
-H "anthropic-version: 2026-10-21" \
-H "content-type: application/json" \
--data '{
"model": "claude-sonnet-5-5",
"max_tokens": 1024,
"messages": [
{"role": "user", "content": "Give me a one-sentence summary of what an API model ID does."}
]
}'
A successful response is a Message object. Its model identifies the model used; content contains content blocks, including text blocks; usage reports input and output token counts; and stop_reason explains why generation ended. For example, end_turn indicates the response finished normally, while max_tokens indicates it reached the requested output limit. See Anthropic’s Messages API example and stop-reasons guide.
For Python, Anthropic’s SDK example uses anthropic.Anthropic().messages.create(...). Install the SDK with pip install anthropic, then pass the model ID, max_tokens, and a user message. See Using the Messages API.
Estimate token costs
Anthropic lists API prices in USD per million tokens (MTok). A first-pass estimate for standard input and output is:
(input tokens ÷ 1,000,000 × input rate) + (output tokens ÷ 1,000,000 × output rate)
For example, assume 100,000 input tokens and 10,000 output tokens, with no prompt caching or Batch API pricing:
- Sonnet 5.5: (100,000 ÷ 1,000,000 × $2) + (10,000 ÷ 1,000,000 × $10) = $0.30.
- Opus 5.5: (100,000 ÷ 1,000,000 × $4) + (10,000 ÷ 1,000,000 × $20) = $0.60.
- Fable 5.1: (100,000 ÷ 1,000,000 × $10) + (10,000 ÷ 1,000,000 × $50) = $1.50.
- Haiku 5.5: for prompts up to 100,000 tokens, the same token counts cost $0.015: $0.01 for input and $0.005 for output. For prompts over 100,000 tokens, the listed rates are $0.50 input and $2.50 output per MTok.
For Haiku 5.5, prompt length includes all input tokens, including cache reads and cache writes; the higher tier applies to a request whose prompt exceeds 100,000 tokens. These are arithmetic estimates, not measured bills. They exclude separately priced cache reads or writes and other applicable features. Anthropic says its Message Batches API reduces costs by 50% for eligible asynchronous requests; see the pricing documentation and Batch processing guide for applicable details.
Check model status before deployment
Check GET /v1/models or Anthropic’s model overview and lifecycle documentation before relying on an ID. Anthropic defines Active as fully supported and recommended, Legacy as no longer updated and potentially subject to deprecation, Deprecated as still functional but no longer recommended, and Retired as unavailable. Requests to retired models fail. Recheck the live inventory when deploying or migrating; an announcement or third-party catalog is not a substitute for the direct API inventory.
Sources
Models overview — Claude Platform Docs
Get your Claude API key — Claude Platform Docs
API overview — Claude Platform Docs
Versions — Claude Platform Docs
Using the Messages API — Claude Platform Docs
Stop reasons and fallback — Claude Platform Docs
Pricing — Claude Platform Docs