Shared inference

Send requests to Morph models without provisioning an endpoint

Call shared models immediately without reserving GPUs. Morph manages the pools and bills usage at each model’s token rates.

Send your first request

Create a dashboard API key. To bill an organization, select it before creating the key.

Set your terminal’s key:

export MORPH_API_KEY="YOUR_MORPH_API_KEY"

These examples use morph-glm53-744b. The catalog lists public model IDs, context windows, and rates.

cURL

curl --fail-with-body https://api.morphllm.com/v1/chat/completions \
  -H "Authorization: Bearer $MORPH_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "morph-glm53-744b",
    "messages": [{"role": "user", "content": "Write a Python function that checks whether a string is a palindrome."}]
  }'

Python

Run pip install openai, then:

import os
from openai import OpenAI

client = OpenAI(
    api_key=os.environ["MORPH_API_KEY"],
    base_url="https://api.morphllm.com/v1",
)

response = client.chat.completions.create(
    model="morph-glm53-744b",
    messages=[{"role": "user", "content": "Write a Python function that checks whether a string is a palindrome."}],
)
print(response.choices[0].message.content)

JavaScript

Run npm install openai. Save as request.mjs; run node request.mjs:

import OpenAI from "openai";

const client = new OpenAI({
  apiKey: process.env.MORPH_API_KEY,
  baseURL: "https://api.morphllm.com/v1",
});

const response = await client.chat.completions.create({
  model: "morph-glm53-744b",
  messages: [{ role: "user", content: "Write a Python function that checks whether a string is a palindrome." }],
});
console.log(response.choices[0].message.content);

Keep keys on your server, never in browser code.

Use your existing client

OpenAI clients need https://api.morphllm.com/v1, your Morph key, and a public model ID.

See Anthropic formats and coding agent setup for other clients.

Handle busy periods and errors

For 429, honor Retry-After when present; retry with exponential backoff and jitter. Immediate retries add load, not capacity.

For 401, check the key and authorization header. For model errors, check the catalog ID. Correct invalid credentials or IDs before retrying.

Standby is a separate tier for work that tolerates capacity rejections; batch processing handles work without interactive responses.

Monitor usage and choose capacity

The dashboard shows requests, usage, and billing for your key’s account. See prompt caching for repeated input.

For reserved capacity, review dedicated plans and cancellation before purchasing. Idle capacity remains billable.