Independent guide. BlackClic is not affiliated with Anthropic and earns no commission from Claude. Specs and prices from Claude’s official documentation and pricing page, November 20, 2026.
Claude Haiku 5.5 is the fastest and lightest model in the family. Anthropic positions it for high-volume, latency-sensitive tasks such as classification, extraction and routing, with API prices starting at $0.10 per million input tokens.

Key specs
| Spec | Claude Haiku 5.5 |
|---|---|
| API ID | claude-haiku-5-5 |
| API price | from $0.1 input / $0.5 output per million tokens |
| Context window | 1 million tokens |
| Max output | 128,000 tokens |
| Speed | Fastest |
| Reliable knowledge cutoff | June 2026 |
| Default effort | medium |
| In the Claude app | All plans, including Free |
Who it’s for
- ✅ Developers building chatbots and customer-support bots
- ✅ Automation builders (n8n, Zapier, Make) processing thousands of items
- ✅ E-commerce stores tagging products and reviews
- ✅ Apps that need instant answers on mobile
When to pick another model
- Complex reasoning or long documents analysis: use Sonnet or Opus
- Large coding projects: Opus 5.5 or Fable 5.2
What 1,000 tasks cost on the API
Example: 1,000 requests of about 2,000 input tokens and 500 output tokens cost from $0.45 with Claude Haiku 5.5. In the Claude app you don’t pay per token: usage is part of your plan.
How it compares to the other Claude models
| Model | Best for | API price (input / output per 1M tokens) | Speed | In the Claude app |
|---|---|---|---|---|
| Fable 5.2 | Hardest reasoning, long agentic projects | $8 / $40 | Moderate | Max & Ultra; Pro with usage credits |
| Opus 5.5 | Long agentic coding and knowledge work | $4 / $20 | Moderate | Pro, Max, Ultra |
| Sonnet 5.5 | Best mix of speed and intelligence | $2 / $10 | Fast | All plans, incl. Free |
| Haiku 5.5 | High-volume, low-latency tasks | from $0.10 / $0.50 | Fastest | All plans, incl. Free |
Next step
Compare plans in our Claude plans and pricing guide, or build your own AI-powered website with this tutorial.
