Skip to main content
The Deepshi API gives you one endpoint for many text models: Deepshi’s own models plus leading third-party models. You choose a model with the model field on each request.
Deepshi is launching with text models (text in, text out). Some models also accept image or file input. See Text models for the full catalogue.

​
Referencing a model

Pass the model’s id in the model field. The id is all you need. 
{ "model": "deepshi-3.0", "messages": [ ... ] }

​
Listing the models your key can use

Call GET /v1/models to see exactly which models your key may call: 
curl https://api.deepshi.ai/v1/models \
  -H "Authorization: Bearer $DEEPSHI_API_KEY"
Each entry’s id is the value you pass as model: 
{
  "object": "list",
  "data": [
    { "id": "deepshi-3.0", "object": "model", "created": 0, "owned_by": "deepshi" },
    { "id": "claude-opus-4.8", "object": "model", "created": 0, "owned_by": "deepshi" }
  ]
}

​
Capabilities

Models advertise the features they support. Common ones:
CapabilityMeaning
toolsFunction / tool calling. See Tool calling.
reasoningEmits a reasoning trace before answering. See Reasoning.
json / structuredSupports response_format JSON and structured outputs. See Structured outputs.
logprobsCan return token log-probabilities.

​
Browse models

Text models

The full catalogue with context windows and capabilities.