Gemini CLI
Google's CLI against the Gemini generateContent API
Gemini CLI calls POST /v1beta/models/{model}:{action}. Point
GOOGLE_GEMINI_BASE_URL at the orientx host and put your key in GEMINI_API_KEY — do not patch
node_modules. The env var is documented in the official
configuration reference.
| Base URL | https://api.orientx.ai |
| Auth | GEMINI_API_KEY (x-goog-api-key) |
| Model | An id from GET /v1/models |
| Official docs | Configuration |
Install
npm install -g @google/gemini-cli
gemini --versionPoint it at orientx
export GEMINI_API_KEY="$ORIENTX_API_KEY"
export GOOGLE_GEMINI_BASE_URL="https://api.orientx.ai"The CLI appends /v1beta/... itself. Leave the /v1 Chat Completions path off this variable.
Run Gemini CLI
geminiOr a one-shot: gemini "Who are you". Switch models with /model; the id must exist on your key.
GOOGLE_GEMINI_BASE_URL must be HTTPS. If a request 404s on /v1/chat/completions, the CLI is
still talking OpenAI — confirm the Gemini env vars are set in that same shell.