claude-haiku-4-5 API
Claude Haiku 4.5 is Anthropic's fastest, most cost-efficient Claude model. Call it on CaMeL Hub through the OpenAI-compatible API (base_url https://api.camel-hub.com/v1) at $1 per 1M input tokens and $5 per 1M output tokens — no separate Anthropic account needed, one API key covers every model on the platform.
Pricing
| per 1M input tokens | per 1M output tokens | |
|---|---|---|
| CaMeL Hub | $1.00 | $5.00 |
| Provider list price | $1.00 | $5.00 |
Prices are the public default-group rate in USD per 1M tokens; input and output are billed separately. · Provider list price · verified 2026-07-11
Specifications
- Provider
- Anthropic
- Context window
- 200000 tokens
- Max output
- 64000 tokens
- Capabilities
- streaming
- Endpoints
- anthropic, openai
What it is good for
- High-volume text classification, intent detection, and content routing where per-call cost dominates
- Real-time chat assistants and autocomplete, where sub-second first-token latency matters more than deep reasoning
- Structured data extraction from long documents — the 200K context takes whole contracts or logs in one request
- Summarizing support tickets, meeting notes, and email threads at scale
- Acting as the low-cost worker model in agent pipelines, executing sub-tasks planned by a stronger model such as Claude Sonnet or Opus
- Bulk offline jobs — tagging, moderation pre-screening, or dataset cleanup across millions of records
Quickstart
OpenAI-compatible: set base_url to https://api.camel-hub.com/v1 and use your CaMeL Hub API key. No SDK changes needed.
cURL
curl https://api.camel-hub.com/v1/chat/completions \
-H "Authorization: Bearer $CAMEL_HUB_KEY" \
-H "Content-Type: application/json" \
-d '{
"model": "claude-haiku-4-5",
"messages": [{"role": "user", "content": "Hello!"}]
}'
Python
from openai import OpenAI
client = OpenAI(
api_key="$CAMEL_HUB_KEY",
base_url="https://api.camel-hub.com/v1", # <-- the only change vs. the OpenAI default
)
resp = client.chat.completions.create(
model="claude-haiku-4-5",
messages=[{"role": "user", "content": "Hello!"}],
)
print(resp.choices[0].message.content)
Node.js
import OpenAI from "openai";
const client = new OpenAI({
apiKey: process.env.CAMEL_HUB_KEY,
baseURL: "https://api.camel-hub.com/v1", // <-- the only change vs. the OpenAI default
});
const resp = await client.chat.completions.create({
model: "claude-haiku-4-5",
messages: [{ role: "user", content: "Hello!" }],
});
console.log(resp.choices[0].message.content);
Get an API key Browse integration guides
Frequently asked questions
How is claude-haiku-4-5 billed on CaMeL Hub?
Usage is metered per token, with input and output priced separately: $1 per 1M input tokens and $5 per 1M output tokens on the public default tier. You only pay for what a request actually consumes — there is no per-call fee and no monthly minimum.
Does it support streaming responses?
Yes. Set stream: true in your request and tokens are delivered incrementally over SSE, exactly as with the OpenAI API. Given Haiku 4.5's fast generation speed, streaming makes chat UIs feel close to instant.
How large are the context window and maximum output?
Claude Haiku 4.5 accepts up to 200,000 tokens of context per request and can generate up to 64,000 output tokens (per Anthropic's published specifications). That is roughly 150,000 English words of input — enough for most long-document workloads without chunking.
How do I start calling it?
Create an API key in the CaMeL Hub console (https://api.camel-hub.com/console), then point any OpenAI SDK at base_url https://api.camel-hub.com/v1 with model "claude-haiku-4-5". Anthropic-native Messages format requests are also accepted, so existing Anthropic SDK code works by swapping the base URL and key.
When should I pick Haiku 4.5 over Claude Sonnet or Opus?
Pick Haiku 4.5 when speed and unit cost matter more than maximum reasoning depth: classification, extraction, summarization, chat, and high-fan-out agent sub-tasks. For complex multi-step reasoning or hard coding tasks, step up to a Sonnet- or Opus-tier model — all available behind the same CaMeL Hub endpoint, so switching is a one-line model-name change.