# claude-haiku-5-5

> LLM model from anthropic, available via Core.Today API.

- **Provider**: anthropic
- **Model ID**: claude-haiku-5-5
- **Category**: chat
- **Pricing Type**: token_based
- **Billing**: per token — see Token Pricing

## API Endpoint

Base URL: https://api.core.today

### Messages
POST /llm/anthropic/v1/messages

## Authentication

Header: `Authorization: Bearer YOUR_API_KEY`

Note: Your Core.Today API key (`cdt_xxx`) works as Bearer token.

## Input Parameters

- `model` (string, **required**): Model identifier. Use `claude-haiku-5-5`
- `messages` (array, **required**): Array of message objects with `role` (`user`/`assistant`) and `content`
- `max_tokens` (integer, **required**): Maximum tokens to generate
- `system` (string, optional): System prompt (a top-level field, not a message role)
- `temperature` (number, optional): Sampling temperature (Range: min: 0, max: 1). Claude Opus 4.7 and later, Sonnet 5 and later, and Fable reject non-default values — omit it for those models
- `stream` (boolean, optional): Enable streaming responses (Default: `false`)

## Example Request

```json
{
  "model": "claude-haiku-5-5",
  "max_tokens": 1024,
  "messages": [
    {
      "role": "user",
      "content": "Hello, how are you?"
    }
  ]
}
```

## Response Format

```json
{
  "id": "msg_abc123",
  "type": "message",
  "role": "assistant",
  "model": "claude-haiku-5-5",
  "content": [
    {
      "type": "text",
      "text": "Response text here"
    }
  ],
  "stop_reason": "end_turn",
  "usage": {
    "input_tokens": 10,
    "output_tokens": 20
  }
}
```

## Usage Flow

1. POST /llm/anthropic/v1/messages with model, max_tokens and messages
2. Response is returned synchronously (or streamed if stream=true)
3. Credits are deducted based on token usage

Credits in the response (opt-in): send the header `X-Coredot-Include-Cost: true` and the response reports the credits that call is billed — `usage.coredot.credits` in the body (Gemini: `usageMetadata.coredot`; streaming: in the final usage event) and, for non-streaming responses, the `X-Coredot-Credits` header. Guide: https://console.core.today/llms/api.txt (section "Credits in the response").

## Token Pricing

- Input: 0.0001858 credits/token (185.8 credits per 1M tokens)
- Cached Input: 1.858e-05 credits/token (18.58 credits per 1M tokens)
- Cache Write: 0.00023225 credits/token (232.25 credits per 1M tokens)
- Image Input: 0.0001858 credits/token (charged at the Input rate)
- Output: 0.000929 credits/token (929 credits per 1M tokens)
- Web Search: 18.58 credits per request (built-in web search tool)

### Long-context tier

Prompts above 100,000 input tokens are billed at these rates for the whole request:

- Input: 0.000929 credits/token (929 credits per 1M tokens)
- Cached Input: 9.29e-05 credits/token (92.9 credits per 1M tokens)
- Cache Write: 0.00116125 credits/token (1,161.25 credits per 1M tokens)
- Output: 0.004645 credits/token (4,645 credits per 1M tokens)
- Web Search: 18.58 credits per request (built-in web search tool)

### Surcharges

When the provider reports a premium tier for a response, every token charge (input, cache, output) is multiplied; per-call fees such as web search are not. Multipliers combine when several apply.

- anthropic_geo_us: x1.1 when response `usage.inference_geo` is `us`

## Image Input

This model accepts images as input. Image bytes are converted to tokens by the upstream provider and charged at the standard Input rate (0.0001858 credits/token).

**Tokenization**: formula-based
- Formula: `tokens = (width * height) / 750`
- Capped at 4784 tokens / 2576px long edge
- Notes: tokens ≈ width × height / 750; Claude 4.7+ (Opus 4.7/4.8/5/5.5, Sonnet 5/5.5, Fable 5/5.1) high-res cap 4784 tokens / 2576px long edge

### Example (Anthropic)

```json
{
  "model": "claude-haiku-5-5",
  "max_tokens": 1024,
  "messages": [
    {
      "role": "user",
      "content": [
        {
          "type": "text",
          "text": "What's in this image?"
        },
        {
          "type": "image",
          "source": {
            "type": "base64",
            "media_type": "image/jpeg",
            "data": "<base64-encoded-image-bytes>"
          }
        }
      ]
    }
  ]
}
```

## Tags

anthropic, chat