Messages

POST https://api.tikway.ai/v1/messages

cURL

Text input

curl --request POST \ --url https://api.tikway.ai/v1/messages \ --header 'Content-Type: application/json' \ --header 'x-api-key: <api-key>' \ --data ' { "model": "claude-sonnet-4-6", "max_tokens": 1024, "system": "You are a helpful assistant.", "messages": [ { "role": "user", "content": "Write a short bedtime story about a unicorn." } ] } '

200

{
  "id": "msg_013Zva2CMHLNnXjNJJKqJ2EF",
  "container": {
    "id": "container_011CpZohnwH4vuy7gazohgSP",
    "expires_at": "2019-12-27T18:11:19.117Z",
    "skills": [
      {
        "skill_id": "pdf",
        "type": "anthropic",
        "version": "latest"
      }
    ]
  },
  "content": [
    {
      "citations": [
        {
          "cited_text": "The grass is green. The sky is blue.",
          "document_index": 0,
          "document_title": "My Document",
          "end_char_index": 0,
          "file_id": "file_011CNha8iCJcU1wXNR6q4V8w",
          "start_char_index": 0,
          "type": "char_location"
        }
      ],
      "text": "Hi! My name is Claude.",
      "type": "text"
    }
  ],
  "context_management": {
    "applied_edits": [
      {
        "cleared_input_tokens": 0,
        "cleared_tool_uses": 0,
        "type": "clear_tool_uses_20250919"
      }
    ]
  },
  "diagnostics": {
    "cache_miss_reason": {
      "cache_missed_input_tokens": 0,
      "type": "model_changed"
    }
  },
  "model": "claude-opus-5",
  "role": "assistant",
  "stop_details": {
    "category": "cyber",
    "explanation": "This request was declined because it conflicts with Anthropic's Usage Policy.",
    "fallback_credit_token": "QW50aHJvcGljL0NsYXVkZQ==",
    "fallback_has_prefill_claim": true,
    "recommended_model": "claude-opus-4-8",
    "type": "refusal"
  },
  "stop_reason": "end_turn",
  "stop_sequence": null,
  "type": "message",
  "usage": {
    "cache_creation": {
      "ephemeral_1h_input_tokens": 0,
      "ephemeral_5m_input_tokens": 0
    },
    "cache_creation_input_tokens": 2051,
    "cache_read_input_tokens": 2051,
    "fallback_credit": {
      "status": {
        "type": "redeemed"
      }
    },
    "inference_geo": "global",
    "input_tokens": 2095,
    "iterations": [
      {
        "cache_creation": {
          "ephemeral_1h_input_tokens": 0,
          "ephemeral_5m_input_tokens": 0
        },
        "cache_creation_input_tokens": 0,
        "cache_read_input_tokens": 0,
        "input_tokens": 0,
        "model": "claude-fable-5-1",
        "output_tokens": 0,
        "type": "message"
      }
    ],
    "output_tokens": 503,
    "output_tokens_details": {
      "thinking_tokens": 0
    },
    "server_tool_use": {
      "web_fetch_requests": 2,
      "web_search_requests": 0
    },
    "service_tier": "standard",
    "speed": "standard"
  },
  "input_transformations": [
    {
      "path": "path",
      "reason": "model_binding_mismatch",
      "type": "thinking_dropped"
    }
  ]
}

Authorizations

x-api-key · string · header · required

Tikway API Key. Passed through the x-api-key request header according to the Anthropic native protocol.

Body

application/json

model · string · required

The Claude model ID used to generate the message, for example claude-opus-5.

max_tokens · integer · required

The maximum number of tokens to generate content. The model can be stopped before reaching the upper limit; set to 0 to only warm up the prompt cache without generating a reply.

Required range: x >= 0

messages · object[] · required

Enter the conversation message. The model works in user and assistant turns; consecutive messages with the same role will be merged. If there is no system role, please use the top-level system for system prompt words.

Maximum array length: 100000

Hide child attributes

messages. role · enum<string> · required

Message role. When the last item is assistant , the model will continue to be generated from its content.

Available options: user, assistant

messages. content · string object[] · required

Message content. A string is equivalent to an array containing only one text block.

system · string object[]

System prompt word. The Messages API does not support system role messages, this field should be used.

stream · boolean

Whether to stream message events as Server-Sent Events.

stop_sequences · string[]

Customize stop sequence; model generation stops when it reaches any sequence.

temperature · number

Sampling temperature. Higher values ​​make the output more random; usually adjusted with top_p, top_k.

Required range: 0 <= x <= 1

top_p · number

Kernel sampling threshold.

Required range: 0 <= x <= 1

top_k · integer

The number of candidate tokens with the highest probability retained during each sampling step.

Required range: x >= 0

thinking · Enabled thinking · object

Extended thinking configuration. Optional enabled, adaptive, disabled, or passed in null. When think is enabled, think tokens count toward max_tokens.

Enabled thinking

Adaptive thinking

Disabled thinking

Hide child attributes

thinking. type · enum<string> · required

Available options: enabled

thinking. budget_tokens · integer · required

Think about the token budget, which must be less than max_tokens.

Required range: x >= 1024

thinking. block_binding · object | null

How to handle when the returned thought block does not pass the dialog prefix check; omitted, null or empty objects use the default behavior.

Hide child attributes

thinking.block_binding. prefix_mismatch_behavior · enum<string> | null · default: error

error will return 400; drop_block will remove unmatched think blocks.

Available options: error, drop_block, null

thinking. display · enum<string> | null · default: summarized

Think about how the content is presented; omitted will hide the content but retain the signature for multiple rounds of conversation.

Available options: summarized, omitted, updates, null

tools · object[]

A tool definition that can be called by Claude.

Hide child attributes

tools. name · string

Tool name.

tools. description · string

Description of tool usage.

tools. input_schema · object

JSON Schema of tool input parameters.

tools. type · string

Tool type; custom tools are usually custom or omitted.

tools. cache_control · object

Hide child attributes

tools.cache_control. type · enum<string>

Available options: ephemeral

tools.cache_control. ttl · enum<string>

Available options: 5m, 1h

tool_choice · object

Controls whether and how Claude selects tools.

Hide child attributes

tool_choice. type · enum<string>

Tool selection mode.

Available options: auto, any, tool, none

tool_choice. name · string

The name of the tool that is forced to be called when type is tool.

tool_choice. disable_parallel_tool_use · boolean

Whether to disable parallel tool calls.

output_config · object

Output configuration.

Hide child attributes

output_config. effort · enum<string>

Output reasoning/quality input level.

Available options: low, medium, high

output_config. format · object

Structured output format.

Hide child attributes

output_config.format. type · enum<string>

Available options: json_schema

output_config.format. schema · object

Target JSON Schema.

output_format · object

JSON output format configuration.

Hide child attributes

output_format. type · enum<string>

Available options: json_schema

output_format. schema · object

Target JSON Schema.

metadata · object

Request metadata.

Hide child attributes

metadata. user_id · string

End user identification.

service_tier · enum<string>

Service level selection.

Available options: auto, standard_only

speed · enum<string>

Inference speed mode.

Available options: standard, fast

container · string object

Code execution container configuration or container ID.

context_management · object

Context management configuration.

mcp_servers · object[]

MCP server definition.

Hide child attributes

mcp_servers. name · string

mcp_servers. url · string

mcp_servers. authorization_token · string

cache_control · object

Request-level prompt word cache control.

Hide child attributes

cache_control. type · enum<string>

Available options: ephemeral

cache_control. ttl · enum<string>

Available options: 5m, 1h

diagnostics · object

Diagnostic information configuration.

fallback_credit_token · string object

Fallback service deduction token.

fallbacks · object

Model fallback configuration.

inference_geo · string

Reasoning about geographic preferences.

Response

200 - application/json

id · string · required

Message unique ID.

type · enum<string> · required

Object type, fixed to message.

Available options: message

role · enum<string> · required

Message role, fixed to assistant.

Available options: assistant

model · string · required

The actual Claude model ID used.

content · object[] · required

List of content blocks generated by Claude.

Hide child attributes

content. type · enum<string> · required

Output content block type.

Available options: text, thinking, redacted_thinking, tool_use, server_tool_use, web_search_tool_result, web_fetch_tool_result, code_execution_tool_result

content. text · string

text Generated text for the content block.

content. citations · object[]

List of textual references.

Hide child attributes

content.citations. type · string

Reference type.

content.citations. cited_text · string

Quoted text.

content.citations. document_index · integer

Source document index.

content.citations. document_title · string

Source document title.

content.citations. start_char_index · integer

content.citations. end_char_index · integer

content.citations. url · string

content.citations. title · string

content. thinking · string

Extended reflection text.

content. signature · string

Think about the content signature; it must remain the same the next time you pass it around.

content. data · string

Encrypted data for redacted_thinking.

content. id · string

Tool call ID.

content. name · string

The name of the called tool.

content. input · object

Tool call input parameters.

content. content · any

Server tool call result content.

content. tool_use_id · string

The associated tool call ID.

content. is_error · boolean

Whether the server tool results are in error.

stop_reason · enum<string> · required

Generate a stop reason.

Available options: end_turn, max_tokens, stop_sequence, tool_use, pause_turn, refusal, model_context_window_exceeded

usage · object · required

Billing and current limiting word usage statistics.

Hide child attributes

usage. input_tokens · integer · required

Enter the number of tokens.

Required range: x >= 0

usage. output_tokens · integer · required

Output the number of tokens.

Required range: x >= 0

usage. cache_creation_input_tokens · integer

The number of input tokens used to create the cache.

Required range: x >= 0

usage. cache_read_input_tokens · integer

The number of input tokens read from the cache.

Required range: x >= 0

usage. cache_creation · object

Hide child attributes

usage.cache_creation. ephemeral_5m_input_tokens · integer

usage.cache_creation. ephemeral_1h_input_tokens · integer

usage. output_tokens_details · object

Hide child attributes

usage.output_tokens_details. thinking_tokens · integer

Think about word count.

usage. server_tool_use · object

Hide child attributes

usage.server_tool_use. web_search_requests · integer

usage.server_tool_use. web_fetch_requests · integer

usage. service_tier · enum<string>

Actual service level.

Available options: standard, priority, batch

usage. speed · enum<string>

Actual reasoning speed.

Available options: standard, fast

stop_sequence · string

If the sequence ends due to a custom stop, it is a hit sequence.

stop_details · object

Structured information for rejection and other stopping situations.

Hide child attributes

stop_details. type · string

stop_details. category · string

stop_details. explanation · string

stop_details. recommended_model · string

container · object

Code execution container information.

Hide child attributes

container. id · string

container. expires_at · string

Container expiration time.

container. skills · string[]

context_management · object

Context management results of server-side applications.

diagnostics · object

Diagnostic information, such as cache miss reasons.

input_transformations · object[]

Transformations applied by the API to the input, such as chunks of thought that are discarded.

Hide child attributes

input_transformations. type · enum<string>

Available options: thinking_dropped

input_transformations. path · string

input_transformations. reason · enum<string>

Available options: model_binding_mismatch, prefix_binding_mismatch, organization_binding_mismatch, end_user_binding_mismatch