Claude Haiku 4.5 API
Use this exact dated snapshot for conversational text, short summaries and response drafting. The protocol and token limits below describe the OneAster route; features available from the original provider are not automatically enabled here.
anthropic/claude-haiku-4-5-20251001 · Anthropic · Text
Availability and pricing
Published contract · ready
$0.96 per 1 million input tokens; $4.8 per 1 million output tokens
Prices shown use the public default price group. Your account price and the request quote determine the amount reserved and charged.
Published verification: 2026-10-05T20:30:50.151Z
Pricing revision: rev_65218e6d-dcfe-467e-8177-49015408a4a0
Supported inputs and limits
messages
- max Input Tokens
- 8192
- max Output Tokens
- 1024
- default Output Tokens
- 256
- tools
- Not enabled
- multimodal Input
- Not enabled
- reasoning
- Not enabled
- cache
- Not enabled
- conversation
- Enabled
- structured Output
- Not enabled
- compact
- Not enabled
- streaming
- Enabled
{
"kind": "messages",
"modes": [
"chat"
],
"qualities": [
"standard"
],
"units": [
1
],
"aspects": [],
"resolutions": [],
"maxReferences": 0,
"maxInputTokens": 8192,
"maxOutputTokens": 1024,
"defaultOutputTokens": 256,
"protocols": [
"messages"
]
}Limitations
- Draft only until live verification and actual cost reconciliation pass. The single cache-write component uses the highest published TTL rate; TTL-specific pricing is not advertised. Tools, vision, explicit reasoning, structured output and compaction are disabled pending dedicated live evidence.
API request example
Use your own API key. This is a request body; asynchronous tasks also require checking their status and output.
POST /v1/messages
{
"model": "anthropic/claude-haiku-4-5-20251001",
"max_tokens": 256,
"messages": [
{
"role": "user",
"content": "Hello"
}
]
}Open API examplesWhich Claude snapshot is this?
Use the exact model ID shown above. The dated snapshot identifier is part of the request; do not substitute an unverified provider alias.
Try this model
Sign in and check your available credits before sending a request.
Open Playground Compare models API documentation