API ReferenceAnthropic

Messages

Use the Anthropic-compatible Messages API when your client expects Anthropic headers and message fields, while Velrix still controls gateway policy, keys, and logs.

Official Anthropic reference

Endpoint

Use the Anthropic-compatible base URL for clients that send x-api-key and anthropic-version headers.

Method

Create an Anthropic-style message.

POST /v1/messages

Base URL

Anthropic-compatible Velrix endpoint.

https://api.velrix.ai

Authentication

Headers

Anthropic's API uses an x-api-key header and an anthropic-version header. Velrix accepts your scoped Velrix key in this provider-compatible shape.

x-api-key

Scoped Velrix API key.

$VELRIX_API_KEY

anthropic-version

Anthropic API version header.

2023-06-01

anthropic-beta

Optional beta feature header for routes that require beta flags.

feature-name

Content-Type

Requests and responses use JSON.

application/json

Schema

Body parameters

These are the primary Anthropic Messages fields. Velrix forwards compatible fields to the selected upstream route when supported.

model

Required

Model ID to route. Use claude-sonnet-4-6 for policy routing or a catalog model ID.

max_tokens

Required

Maximum number of tokens to generate. Anthropic requires this field for Messages requests.

messages

Required

Conversation messages using Anthropic-style user and assistant roles.

system

Optional

System prompt. Anthropic documents system as a top-level parameter, not a message role.

metadata

Optional

Attach request metadata, including user identifiers or operational context when supported.

service_tier

Optional

Select the requested service tier for latency/capacity behavior when supported by the route.

stop_sequences

Optional

Custom sequences that cause generation to stop when encountered.

stream

Optional

Return incremental server-sent events when set to true.

temperature / top_p / top_k

Optional

Sampling controls. Use one primary randomness control for predictable production behavior.

thinking

Optional

Configure extended thinking behavior for Anthropic models that support it.

tools / tool_choice

Optional

Define client tools and control whether tool use is automatic or constrained.

Request

Request example

Send Anthropic-style messages with max_tokens and the required version header.

Shell

curl https://api.velrix.ai/v1/messages \
  -H "x-api-key: $VELRIX_API_KEY" \
  -H "anthropic-version: 2023-06-01" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "claude-sonnet-4-6",
    "max_tokens": 512,
    "system": "Answer with concise operational detail.",
    "metadata": {
      "user_id": "dashboard-docs"
    },
    "messages": [
      {
        "role": "user",
        "content": "Summarize the failed deployment."
      }
    ],
    "temperature": 0.2
  }'

TypeScript

import Anthropic from "@anthropic-ai/sdk";

const client = new Anthropic({
  apiKey: process.env.VELRIX_API_KEY,
  baseURL: "https://api.velrix.ai",
});

const message = await client.messages.create({
  model: "claude-sonnet-4-6",
  max_tokens: 512,
  system: "Answer with concise operational detail.",
  messages: [
    {
      role: "user",
      content: "Summarize the failed deployment.",
    },
  ],
});

console.log(message.content);

TypeScript stream

import Anthropic from "@anthropic-ai/sdk";

const client = new Anthropic({
  apiKey: process.env.VELRIX_API_KEY,
  baseURL: "https://api.velrix.ai",
});

const stream = await client.messages.create({
  model: "claude-sonnet-4-6",
  max_tokens: 512,
  stream: true,
  messages: [{ role: "user", content: "Write a rollout note." }],
});

for await (const event of stream) {
  console.log(event.type);
}

Response

Response 200

Messages responses include content blocks and usage metadata in the Anthropic-compatible shape.

Response 200

{
  "id": "msg_013Zva2CMHLNnXjNJJKqJ2EF",
  "container": {
    "id": "id",
    "expires_at": "2019-12-27T18:11:19.117Z"
  },
  "content": [
    {
      "citations": [
        {
          "cited_text": "cited_text",
          "document_index": 0,
          "document_title": "document_title",
          "end_char_index": 0,
          "file_id": "file_id",
          "start_char_index": 0,
          "type": "char_location"
        }
      ],
      "text": "Hi! My name is Claude.",
      "type": "text"
    }
  ],
  "model": "claude-opus-4-6",
  "role": "assistant",
  "stop_details": {
    "category": "cyber",
    "explanation": "explanation",
    "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,
    "inference_geo": "inference_geo",
    "input_tokens": 2095,
    "output_tokens": 503,
    "server_tool_use": {
      "web_fetch_requests": 2,
      "web_search_requests": 0
    },
    "service_tier": "standard"
  }
}

Routing

Routing notes

Anthropic-compatible requests use the same Velrix operational model as OpenAI-compatible requests.

Native message schema

Keep Anthropic-style client code on its expected fields.

Scoped keys

Reuse Velrix key scopes, quotas, and environment separation.

Policy routing

Route by model policy without changing Anthropic-compatible clients.

Streaming

Use stream: true for server-sent event streaming when the selected route supports it.