CLI Reference¶

All inferencehub commands. Run inferencehub <command> --help for all flags.

Commands¶

Command	Description
`inferencehub config init`	Generate a starter `inferencehub.yaml` in the current directory
`inferencehub config validate --config <file>`	Validate config and check required env vars
`inferencehub config show --config <file>`	Show config after env var interpolation
`inferencehub install --config <file>`	Install the platform onto the cluster
`inferencehub upgrade --config <file>`	Upgrade an existing installation
`inferencehub status`	Show component health
`inferencehub verify`	Check prerequisites and platform readiness
`inferencehub uninstall --confirm`	Remove the platform

Supported model providers¶

Provider	Example model ID
AWS Bedrock	`anthropic.claude-3-5-sonnet-20241022-v2:0`
OpenAI	`gpt-4o`
Ollama (self-hosted)	`llama3.2:3b`
Azure OpenAI	`gpt-4`

Working directory¶

Run all inferencehub commands from the project root — the directory that contains helm/ and scripts/. The CLI reads inferencehub.yaml and .env from the current directory.

cd /path/to/inference-hub
inferencehub install --config inferencehub.yaml

Environment variables¶

The CLI auto-loads env files in this order:

.env (project root)
.env.local (project root)
~/.inferencehub/.env (user global)

Variable	Required	Description
`LITELLM_MASTER_KEY`	Yes	Must start with `sk-`. Used as the LiteLLM API master key.
`LANGFUSE_PUBLIC_KEY`	No	Required only when `observability.enabled: true`
`LANGFUSE_SECRET_KEY`	No	Required only when `observability.enabled: true`