Models
Fetch every available model and its price over the API, or read the notes below.
List models API
GET https://api.eclogue.ai/v1/models {
"object": "list",
"data": [{
"id": "deepseek-chat",
"object": "model",
"owned_by": "deepseek",
"context_length": 128000,
"pricing": {
"prompt": "0.002",
"completion": "0.008",
"cached_prompt": "0.0005"
},
"input_modalities": ["text"],
"output_modalities": ["text"],
"model_card": {
"description": "…",
"reasoning": true,
"tool_call": true,
"attachment": false,
"max_output": 8192
},
"capabilities": {"streaming": true, "function_calling": true},
"status": "active"
}]
} Text models
Covers the major model families (Claude, Gemini, DeepSeek, Qwen, GLM and more). Exact model names, prices and context lengths change over time — check them via:
GET /v1/modelsfor the live model list- the pricing page for current prices across all models
Video models
Video generation models such as Seedance are supported. See the video generation guide.
Metadata fields
Read the model metadata fields in the response as follows:
input_modalities/output_modalities— media types the model accepts and produces, drawn fromtext,image,audio,videoandpdf.nullmeans the metadata has not been recorded yet, not that the type is unsupported; an empty array means no additional modality was confirmed.model_card— description and capability metadata (description,reasoning,tool_call,attachment,max_output). Here too,nullmeans "not recorded" rather than "not available".capabilities— call-level flags, always an object. Itsvisionis derived from whetherinput_modalitiescontainsimage, andfunction_callingfrommodel_card.tool_call. New integrations should read those two sources directly — their semantics are more explicit.
Model naming
Unlike OpenRouter's provider/model convention, this gateway takes a plain model ID (e.g. deepseek-chat) and routes it to the best available node for you.