OpenAI compatibility
ModelDock exposes your launched models through the OpenAI chat completions API. It also accepts the Anthropic Messages format, so Claude Code and the Anthropic SDKs can point at it. The base URL for the OpenAI SDKs is https://api.lab.litepod.io/v1.
Endpoints
| Endpoint | What it does |
|---|---|
GET /v1/models | List your active launched models, with context length and per-million-token prices. |
GET /v1/models/{id} | Retrieve one of your models. |
POST /v1/chat/completions | OpenAI-compatible chat completion, with optional streaming. |
POST /v1/messages | Anthropic-compatible message. Use the base URL https://api.lab.litepod.io for the Anthropic SDKs. |
POST /v1/messages/count_tokens | Estimated prompt size for a messages request. Free. |
There are no embeddings, legacy completions, images or audio endpoints.
Chat completion fields
ModelDock validates only the fields its limits depend on and forwards the rest to the model server as you sent them:
| Field | Behaviour |
|---|---|
model | Required. The id of one of your launched models. |
messages | Required. The conversation so far. |
temperature | 0 to 2. |
max_tokens / max_completion_tokens | Send one, not both. Must not exceed the model's context length. |
n | 1 to 128. Pre-flights allow n=1. |
stream | true for server-sent events. See Streaming. |
| Anything else | Forwarded as is. Whether the model honours it depends on the model and its server. |
Because extra fields are forwarded, features such as tool calling or image input work only if the model you launched supports them. Check the model card on Hugging Face.
Model ids
Each launch gets its own id, modeldock/<model>-<suffix>. The responses to GET /v1/models add context_length and a pricing object (input and output USD per million tokens) to the usual OpenAI fields.
curl https://api.lab.litepod.io/v1/models -H "Authorization: Bearer $MODELDOCK_API_KEY"More from litepod: the free LLM API with free open-source models. ModelDock is a litepod product.