Claude Sonnet 5.5 — HopsAPI Docs

Claude Sonnet 5.5 is a chat model in the HopsAPI catalog. This page documents only gateway-verified protocols, endpoints, and model IDs; context length, vision, tools, reasoning, and JSON capabilities depend on the current upstream and live response.

Model information

Item	Value
Protocol	Anthropic Messages
Channel	Subscription Direct (`flex`) only
Endpoint	`POST /v1/messages`
Availability	Use the current API key's `GET /v1/models` response
Pricing	$2 input / $10 output per 1M tokens, before your account coefficient

Model IDs

Model ID	Notes
`claude-sonnet-5-5`	Use this ID when returned by the current API key's model list

Do not assume that the model is enabled for your account from this page alone. Before calling it, query GET /v1/models and use the returned model ID, endpoint, and optional responses_endpoint exactly. Official API Relay (standard) keys do not list this model.

Official prices

Prices are USD per 1M tokens, before your account coefficient. There is no long-context surcharge.

Item	Price
Input	$2.00
Output	$10.00
Cache read	$0.20
5-minute cache write	$2.50
1-hour cache write	$4.00

Cache read is 0.1× the input price. Anthropic reports cache writes in usage.cache_creation_input_tokens. The 1-hour subset is usage.cache_creation.ephemeral_1h_input_tokens and uses the $4.00 rate; the remainder uses the 5-minute cache-write price. See Billing & reconciliation.

Core request parameters

Parameter	Required	Description
`model`	Yes	Use an ID above that is returned by the model list
`messages`	Yes	Message array in the protocol's native format
`max_tokens`	Yes	Limits model output length
`temperature`	No	Sampling parameter; accepted range depends on the current upstream
`stream`	No	Use the protocol-specific streaming endpoint or field

Request example

curl https://hopsapi.com/v1/messages \
  -H "x-api-key: sk-YOUR_KEY" \
  -H "anthropic-version: 2023-06-01" \
  -H "content-type: application/json" \
  -d '{"model":"claude-sonnet-5-5","max_tokens":512,
       "messages":[{"role":"user","content":"Introduce this model in one sentence."}]}'

Response example

This example shows the protocol shape only. Text, token counts, and additional fields such as id and model come from the live response.

{
  "type": "message",
  "role": "assistant",
  "content": [
    { "type": "text", "text": "This chat model is called through the Anthropic Messages protocol." }
  ],
  "stop_reason": "end_turn",
  "usage": { "input_tokens": 18, "output_tokens": 22 }
}

Request notes

These constraints apply to claude-sonnet-5-5:

Do not send thinking with type set to disabled. The model rejects it.
tool_choice accepts auto and none. Values any and tool are rejected. The same restriction applies when an OpenAI request uses tool_choice set to required or a named function.
Do not prefill an assistant message. The model rejects assistant prefill.
A thinking block belongs to the model and conversation that produced it. Start a new conversation when switching models.

The gateway passes through native protocol responses and streaming events. See Anthropic format.