Developers

Merkium is a surface for your software too. If your product speaks MCP, it can sit inside every Merkium conversation.

Integrate your app over MCP

Expose your product as tools on a Model Context Protocol server and Merkium can call them mid-conversation. Once approved, your integration appears in the store.

Ship it from the Merkium Console

Register integrations, publish skills, write the tool descriptions users see, and track each submission through review.

An OpenAI-compatible API

Point any OpenAI client at https://chat.merkium.nl/api/v1, or send plain HTTP. Keys live in the console, five per account, each with $5.00 of free credits behind it. Every model is free today, so input and output cost $0.00. Streaming, reasoning, and tool calls all come through unchanged.

curl
curl https://chat.merkium.nl/api/v1/chat/completions \
  -H "Authorization: Bearer $MERKIUM_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "ganymede-1.1",
    "messages": [{"role": "user", "content": "Explain KV caching."}]
  }'
OpenAI SDK
from openai import OpenAI

client = OpenAI(
    base_url="https://chat.merkium.nl/api/v1",
    api_key=os.environ["MERKIUM_API_KEY"],
)
reply = client.chat.completions.create(
    model="ganymede-1.1",
    messages=[{"role": "user", "content": "Explain KV caching."}],
)
print(reply.choices[0].message.content)
ModelGood for
ganymede-1.1Ultra. 900B-class, 1M context.
pluto-1.2-proMax tier. Deep reasoning.
pluto-1.2-flashFast, with a quick reason.
pluto-1.1-flashCode and security, small.
pluto-1.1-proPro reasoning and security scripting.
ceres-1.0Everyday assistant.
ceres-1.0-proMost capable general model.

List everything the API serves with GET https://chat.merkium.nl/api/v1/models. Manage your keys

Ready when you are: the Console