Models

simo-1-pro is the default in the SDK and the dashboard. Requests for simo-1 are refused with model_not_available until it opens.

#Simo-1 Pro

Simo-1 Pro reads text, images and video and answers all six question types.

• Fast when it is sure. On a confident decision it answers at full speed.
• Careful when it is not. When a decision is genuinely uncertain, it deliberates before answering. You don't turn this on or pay for a separate mode; it happens only where it helps.
• Calibrated. Decisions come with probabilities that match how often they are right, so thresholds in your code mean what they say.
• Multimodal. Text, JSON states, screenshots, photos and video clips, in one request.

#Simo-1

Simo-1 is built for real-time judgments, answering every decision in a single pass. It isn't open yet. GET /v1/models (or c.models()) reports when it becomes available.

#Choosing a model

Use simo-1-pro. It is the only model available today, and its speed on confident decisions makes it suitable for loops as well as for expensive calls.

c = Client()                          # simo-1-pro by default
c.ask(state, questions, model="simo-1-pro")

{"model": "simo-1-pro", "state": "...", "questions": {...}}

#List models

curl https://api.temprl.org/v1/models

c.models()   # [{"id": "simo-1-pro", "available": True, ...}, {"id": "simo-1", "available": False, ...}]

{
  "object": "list",
  "models": [
    {"id": "simo-1-pro", "name": "Simo-1 Pro", "available": true, "status": "available"},
    {"id": "simo-1", "name": "Simo-1", "available": false, "status": "coming_soon"}
  ]
}