Skip to Content
We are live but in Staging 🎉
ModelsAPI ReferenceModels (list & get)

Models (list & get) — API Reference

Package: dodil.ignite.v1 · Service: ModelService

Discover the catalog and inspect a single model’s metadata. The HTTP surface is OpenAI-compatible: the paths are the top-level /v1/models and /v1/models/{model} (not under /v1/ignite/).

Both RPCs are public — no auth header required. A model’s declared input.format tells the client which RPC to call (chat, embeddings, rerank, transcription, or generic infer). For the human-readable list, see the Model Catalog; to use an OpenAI/Cohere client, see Using OpenAI & Cohere SDKs.

RPCHTTPstreaming
ListModelsGET /v1/modelsunary (no auth)
GetModelGET /v1/models/{model}unary (no auth)

gRPC reaches both methods at dodil.ignite.v1.ModelService/<Method> on $IGNITE_GRPC. See Conventions → Using gRPC for grpcurl setup.

ListModels

Lists the catalog. Optional filters narrow by family, input_modality, or tags.

Request

# Public endpoint — no Authorization header curl -sS "https://api.dev.dodil.io/v1/models?input_modality=text&family=qwen"

Response

{ "object": "list", "data": [ { "id": "kimi-k2.5", "object": "model", "created": 1748390400, "owned_by": "dodil", "name": "Kimi K2.5", "family": "kimi", "input_modality": "text", "output_modality": "text", "task": "chat", "billing_mode": "token", "context_window": 256000, "input": { "format": "chat" }, "output": { "format": "chat" } } ] }

GetModel

Returns the full ModelInfo for one model id.

Request

# Public endpoint — no Authorization header curl -sS "https://api.dev.dodil.io/v1/models/kimi-k2.5"

Response

{ "data": { "id": "kimi-k2.5", "object": "model", "created": 1748390400, "owned_by": "dodil", "name": "Kimi K2.5", "description": "Long-context chat and reasoning model.", "family": "kimi", "architecture": "mixture-of-experts", "input_modality": "text", "output_modality": "text", "task": "chat", "tags": ["chat", "reasoning", "long-context"], "category": ["text-generation"], "billing_mode": "token", "parameters": "1T", "context_window": 256000, "quantization": "fp8", "precision": "bf16", "languages": ["en", "zh"], "performance": "high", "use_cases": ["assistants", "agentic-tools"], "input": { "format": "chat" }, "output": { "format": "chat" } } }

ModelInputSpec.format tells the client which RPC to call for this model — chat, embeddings, rerank, transcription, or a generic infer shape. For the curated, human-readable listing, see the Model Catalog.


See also