# Model Governance

Learn how PHOENIQS names its models, tests updates, provides long-term support, and manages models throughout their lifecycle.

# How model naming works

PHOENIQS uses semantic versioning principles to manage AI model versions, providing stable model identifiers while allowing the underlying models to evolve. Each supported major version has a consistent model name. For example, an application using GLM 5 can continue using the same model identifier when PHOENIQS introduces a validated minor update, such as GLM 5.4. This approach provides continuity for applications and reduces the need for configuration changes as models are updated.

Identifier (illustrative)	Meaning	Production guidance
`GLM 5`	Supported major-version LTS identifier	Use for production where listed as Active
`GLM 5.4-preview`	Minor update being tested	Experimental; do not use for production
`GLM 6`	New major family, separately evaluated and introduced	Use only when listed as Active

These names are examples. Check the Active Models page for available models and their exact API identifiers.

# In this section

Model naming: how version identifiers work and stay stable.

model-naming/

Model lifecycle: catalog statuses from preview to decommissioned.

model-lifecycle/

Minor updates: how preview, validation, and promotion work.

minor-updates/

Retirement and migration: notices, timelines, and on-demand status.

retirement-and-migration/

Custom model hosting: continue using a model beyond standard retirement.

custom-hosting/

API discovery: find available model identifiers through the API.

api-discovery/

Frequently asked questions about versioning and lifecycle.

faq/

maas model governance lifecycle versioning