Models

All model calls use Vichar model IDs that start with vichar-, for example vichar-sonnet-5.5. One ID maps to one underlying model; the gateway picks a provider serving it.

Get the live list

• Web: vichar.io/models
• API: GET https://api.vichar.io/v1/models

curl https://api.vichar.io/v1/models -H "Authorization: Bearer $VICHAR_API_KEY"

Each entry includes the display name, context window, max output, per-token pricing in USD, and supported parameters. Paid models bill at the listed rates; models marked experimental may have variable availability and rate limits.

auto and smart routing

Instead of a model ID you can pass auto (or smart) and the gateway classifies the request by difficulty and picks from your configured candidate list. If your organization has not configured one, the default tier list is used: an Opus-class model for hard requests, Sonnet-class for medium, Haiku-class for easy, and a cost-efficient model for the simplest requests. Configure the list and classifier under Organization settings > Smart Routing in the dashboard.

curl https://api.vichar.io/v1/chat/completions \
  -H "Authorization: Bearer $VICHAR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"model":"auto","messages":[{"role":"user","content":"Hi"}]}'

The response metadata shows which model was selected.

Experimental models

IDs for experimental models are listed in the catalogue with free: true. They are useful for testing; availability, context length, and rate limits vary, so do not depend on them for production traffic.