Concepts

Model aliases: fast, balanced and best

SubToAPI exposes three stable Claude tiers instead of raw model ids. What each alias is for, how to pick one, and how to list them with GET /v1/models.

Updated

Your code names a *tier*, not a dated model id. When the provider ships a better model for that tier the deployment is updated server-side and your integration keeps working unchanged.

Aliases
fast
lowest latency
Classification, routing, extraction, short replies, high-volume background jobs. No extended thinking.
balanced
default
Most application traffic: assistants, summaries, drafting, tool use. Supports thinking budgets.
best
highest capability
Hard reasoning, long documents, agentic multi-step tool use. Supports thinking budgets. Slowest and most token-hungry.

Listing models

terminal
curl "https://api.subtoapi.app/v1/models" -H "Authorization: Bearer $SUBTOAPI_KEY"
200 OK
{
  "models": [
    { "id": "fast", "name": "Fast", "provider": "claude", "description": "Lowest-latency Claude tier configured for this deployment." },
    { "id": "balanced", "name": "Balanced", "provider": "claude", "description": "Default Claude tier for most application requests." },
    { "id": "best", "name": "Best", "provider": "claude", "description": "Highest-capability Claude tier configured for this deployment." }
  ],
  "request_id": "req_…"
}

Choosing a tier

  • Start on balanced. Move individual call sites to fast once you know the task is simple.
  • Reserve best for the few requests where quality matters more than latency — and measure with the usage page.
  • Tier names are accepted in model on every endpoint and echoed back in the response.

Frequently asked questions

Can I pin a specific Claude model id?
No. Tiers are deliberately stable aliases so that provider model updates never break your integration.
Do tiers affect price?
SubToAPI charges for the platform, not per token. Your Claude usage remains subject to the provider access authorized for the deployment.