Skip to content
Pawan.KrdBeta
API ReferenceAnthropic Messages

API Reference

Anthropic Messages API

An Anthropic-compatible /v1/messages endpoint for every chat model on the platform, so the Anthropic SDKs, Claude Code and other Anthropic clients work by changing only the base URL and key. Authenticate with x-api-key (or Authorization: Bearer). Pricing, plan limits and model access are the same as Chat Completions; errors use Anthropic's error format.

POST/v1/messages

Quick example

curl
curl https://api.pawan.krd/v1/messages \ -H "x-api-key: pk-your_key" \ -H "anthropic-version: 2023-06-01" \ -H "content-type: application/json" \ -d '{ "model": "pkrd/cosmosrp-3.5", "max_tokens": 1024, "system": "You are a helpful assistant.", "messages": [{ "role": "user", "content": "Hello!" }] }'
Python (Anthropic SDK)
import anthropic client = anthropic.Anthropic(base_url="https://api.pawan.krd", api_key="pk-your_key") message = client.messages.create( model="pkrd/cosmosrp-3.5", max_tokens=1024, messages=[{"role": "user", "content": "Write a haiku about Erbil."}], ) print(message.content[0].text)

Parameters

modelstringrequired
Any chat model ID on the platform.
max_tokensintegerrequired
Maximum tokens to generate.
messagesarrayrequired
User / assistant turns. Content blocks: text, image (base64 or url), document (text or base64 PDF), tool_use, tool_result.
systemstring | array
System prompt (text blocks; cache_control is accepted and ignored).
toolsarray
Client tools with input_schema. Anthropic server tools (web search, code execution, computer use, …) are not available and are ignored.
tool_choiceobject
auto, any, tool or none; disable_parallel_tool_use is honoured.
thinkingobject
{ type: "enabled", budget_tokens } turns on reasoning for reasoning models and returns thinking blocks.
streamboolean
Anthropic server-sent events (message_start … message_stop).
stop_sequencesarray
Up to 4 stop sequences.
temperature / top_p / top_knumber
Sampling controls, clamped to the model's ranges.
metadata.user_idstring
End-user identifier.

Tool use

The model replies with tool_use blocks and stop_reason: "tool_use". Send the result back in the next user message as a tool_result block with the same tool_use_id.

Request with a tool
{ "model": "openai/gpt-oss-20b", "max_tokens": 1024, "tools": [ { "name": "get_weather", "description": "Current weather for a city", "input_schema": { "type": "object", "properties": { "city": { "type": "string" } }, "required": ["city"] } } ], "messages": [{ "role": "user", "content": "Weather in Duhok?" }] }

Other endpoints

POST/v1/messages/count_tokens
GET/v1/models

count_tokens returns { "input_tokens": N } and is free. /v1/models answers in Anthropic's format when the anthropic-version header is sent.

Claude Code
See Coding Agents to run Claude Code against any model on Pawan.Krd.