Skip navigation
Contents

Getting started

Models & pricing

The models currently on sale and their list prices, live from the platform. Prices are in USD per 1 million (1M) tokens.

Available models#

  • Type: Chat models are used with /v1/chat/completions, Embedding models with /v1/embeddings. Calling the wrong endpoint → 400 model_type_mismatch.
  • Input: prompt tokens (messages, tools…). Output: tokens the model generates.
  • Cache read: the price for prompt tokens served from the provider’s cache (if the model supports it). Blank = not applicable.
  • Max output: the model’s max_tokens limit. Larger values are clamped to it.

Your price may differ

The table shows list prices. Accounts on a custom price plan (enterprise, resellers) may have different prices — sign in and open Models & pricing to see the prices that apply to you.

List models via the API#

GET /v1/models returns the models your API key may call (OpenAI format, without prices); each model has an extra kind field (chat | embedding). Use the id value as the model parameter.

GET /v1/models
curl https://<host>/v1/models \
  -H "Authorization: Bearer sk-kompa-YOUR_API_KEY"

Model names#

You always call a model by its Any name. Behind the scenes, a model may be served by several providers for load balancing and failover; the response always returns the model name you called and never reveals the provider.

For per-token billing, see Billing.

The pricing table above comes from GET /api/public/models (no authentication): each model has a list_price_usd_per_1m with input, output, cache_read and cache_write — exact USD decimal strings per 1M tokens (e.g. "0.15").