API index
Call catalog models with credits. Not every text API is OpenAI SDK compatible.
Inference Base URL is https://apirouter.pleum.ai/v1. Anthropic and Gemini SDKs use the root origin. https://apirouter.pleum.ai
Operations#
| method | path | auth | summary |
|---|---|---|---|
| POST | /v1/chat/completions | API key (plm_) | Chat completions (OpenAI-compatible) |
| POST | /v1/messages | API key (plm_) | Messages (Anthropic-compatible) |
| POST | /v1beta/models/{model}:generateContent | API key (plm_) | Gemini generateContent |
| POST | /v1/responses | API key (plm_) | Responses (OpenAI) |
| POST | /v1/responses/compact | API key (plm_) | Conversation compact (unsupported · 501) |
| POST | /v1/images/generations | API key (plm_) | Image generation |
| POST | /v1/images/edits | API key (plm_) | Image editing |
| WS | /v1/realtime | API key (plm_) | Realtime voice |
| POST | /v1/audio/speech | API key (plm_) | TTS |
| POST | /v1/audio/generations | API key (plm_) | Music generation (currently unavailable) |
| POST | /v1/video/generations | API key (plm_) | Video generation |
| POST | /v1/3d/generations | API key (plm_) | 3D generation |
| POST | /v1/embeddings | API key (plm_) | Embeddings |
| POST | /v1/rerank | API key (plm_) | Rerank (currently unavailable) |
| POST | /v1/systemone | API key (plm_) | System One (decisions) |
| GET | /v1/models | Public | Model list (public) |
| GET | /v1/credits | JWT or API key | Credit balance |
| GET | /v1/generation | JWT or API key | Single request lookup (X-Request-Id) |
| GET | /v1/fx-rate | Public | Display FX rate |
| GET | /v1/keys | Dashboard JWT | List API keys |
| POST | /v1/provisioning/keys | Provisioning key (plmk_) | Mint keys via provisioning |
| GET | /v1/routing-policies | Dashboard JWT | Routing policies |
| GET | /v1/usage/report | API key (plm_) | Usage report (usage:read key) |
| POST | /v1/auth/device/code | Public | Device login code |
| GET | /v1/stats/models | Public | Public model stats |
| GET | /v1/stats/apps | Public | App attribution stats |
Where to find cost#
Cost is settled after the reply. Find it in one of the four places below; each API doc links to this table.
| Where | Applies to | Note |
|---|---|---|
cost (response body) | Non-streaming chat, embeddings, images, and transcription, plus System One | May be omitted when settlement is late. |
X-Cost-Krw / X-Cost-Usd (response headers) | messages, responses, gemini, and text-to-speech (TTS). Not on streams | Values are decimals in 0.1 steps. |
cost / ttft_ms (last stream chunk) | Streaming chat | Included only once settlement has finished. |
GET /v1/generation?id= | Every request. id is the X-Request-Id response header value | The most reliable lookup. Returns your own requests only. |
See the error codes doc for response shapes.