Skip to main content

GoingMerry Cloud

GoingMerry Cloud allows you to seamlessly run massive, high-performance model configurations (such as Qwen 4 480B or Llama 4 405B) that exceed local consumer hardware memory bounds by offloading inference to secure, managed GPU infrastructure.

Cloud execution integrates natively with the merry CLI and local API ports. When executing a cloud model, input processing and token streaming behave identically to local endpoints, enabling direct compatibility with all local developer tools, IDE extensions, and automated coding agents.


1. Authentication

To execute cloud models, you must have an active GoingMerry account.

Authenticate your terminal shell by executing:

merry signin

This will output a validation code and open a browser window to securely authorize your local daemon environment.


2. Running Cloud Models

Once signed in, you can run, pull, or query cloud-enabled models by adding the -cloud or :cloud suffix:

merry run qwen4-coder:480b-cloud

Supported API Endpoints

All cloud-suffix models can be programmatically queried via the standard local REST endpoints:

curl http://localhost:11434/api/generate -d '{
"model": "qwen4-coder:480b-cloud",
"prompt": "Write a high-performance vector collision function in Rust.",
"stream": false
}'

3. Data Privacy and Security

GoingMerry is engineered with a strict air-gapped philosophy. When utilizing GoingMerry Cloud:

  • Zero Retention: Input prompts and output tokens are processed in volatile memory. No user transaction data, prompts, or weights are written to persistent storage in the cloud.
  • Encryption in Transit: All communications between your local merry daemon and GoingMerry Cloud endpoints are protected using TLS 1.3.

4. Plans & Quotas

Access to GoingMerry Cloud is divided into three subscription tiers:

TierConcurrent Cloud ModelsPricingBest For
Free1FreePersonal projects & testing
Pro3$20 / moProfessional developers
Max10$100 / moTeams and heavy agentic workloads

For enterprise integrations featuring Single Sign-On (SSO) Okta/Azure AD configurations, contact [email protected].