Claude Haiku 4.5 API pricing
Claude Haiku 4.5 costs $1 per million input tokens and $5 per million output tokens. A typical chatbot request (about 1,500 input + 400 output tokens) costs $0.0037, or $3.65 per 1,000 requests.
0 characters0 words0 tokens (GPT o200k)
| Model | Per month |
|---|---|
| Claude Haiku 4.5 |
≈ = estimated token count: only OpenAI publishes its tokenizer. Claude counts use Anthropic’s published ratios; use the count_tokens API for an exact number.
Claude Haiku 4.5 price per million tokens
USD, Anthropic standard tier. Prices checked on October 28, 2026.
| Token type | Price per 1M | With Batch API |
|---|---|---|
| Input | $1 | $0.5 |
| Cached input | $0.1 | $0.05 |
| Cache write (5 min) | $1.25 | $0.625 |
| Output | $5 | $2.5 |
Model ID: claude-haiku-4-5 · Source: platform.claude.com
What a Claude Haiku 4.5 request costs
| Request | Per request | Per 1,000 |
|---|---|---|
| Short chat reply 500 input + 200 output tokens | $0.0015 | $1.55 |
| Chatbot turn with context 1,500 input + 400 output tokens | $0.0037 | $3.65 |
| Document Q&A (RAG) 8,000 input + 500 output tokens | $0.01 | $11.30 |
| Long-document summary 150,000 input + 2,000 output tokens | $0.18 | $175 |
Token counts are measured with GPT's o200k tokenizer and converted for Claude Haiku 4.5 (see below). No caching or batch discount.
Similarly priced models
| Model | Chatbot request | vs Claude Haiku 4.5 |
|---|---|---|
| GPT-5.4 mini OpenAI | $0.0029 | −20% |
| Grok 4.3 xAI | $0.0030 | −19% |
| Grok 4.20 xAI | $0.0030 | −19% |
| o4-mini OpenAI | $0.0034 | −7% |
| GPT-6.1 Terra OpenAI | $0.0035 | −4% |
| DeepSeek V4 Pro DeepSeek | $0.0037 | same |
Frequently asked questions
How much does the Claude Haiku 4.5 API cost?
Claude Haiku 4.5 costs $1 per million input tokens and $5 per million output tokens on Anthropic's standard tier. Prices checked on October 28, 2026.
How much does one Claude Haiku 4.5 request cost?
A typical chatbot request with about 1,500 input and 400 output tokens costs about $0.0037, or $3.65 per 1,000 requests. Paste your own prompt into the calculator on this page for an exact estimate.
Does Claude Haiku 4.5 support prompt caching?
Yes. Cached input costs $0.1 per million tokens, 90% less than regular input. Writing to the cache (5-minute lifetime) costs $1.25 per million tokens.
Does Claude Haiku 4.5 have a Batch API discount?
Yes: 50% off for requests sent through the Batch API, which may take up to 24 hours. That makes it $0.5 / $2.5 per million input / output tokens.
How are Claude Haiku 4.5 tokens counted?
Claude Haiku 4.5 uses Anthropic's earlier tokenizer. We estimate about 1.10× the GPT o200k count for the same text.