Use our smart auto model when you want it cost-effective. Pick a frontier (BYOK) or open model when you want that quality every time.
Same catalog, multiple APIs. You can use your keys, or buy credits for our open models.
Send model: "kasva/auto". We look at the prompt and pick a route: cheap for short chat, a coder for code, a stronger model when it looks hard. Your keys first, then our open models.
OpenAI-compatiblehttps://kasva.dev/v1Anthropic-compatiblehttps://kasva.dev
Paste a key from the provider you already use. You stay on their bill. We route. We don’t resell their models.
No credits, they stay off the list. Add credits, they show up.
If a provider is down or rate-limited, we hop to the next one. Your keys first, then our open models.
Our image endpoints support various image models.
Our video endpoints support various video models.
Key, URL, then your provider key or credits.
Register, copy the kv_… key once. Shown once.
OpenAI: https://kasva.dev/v1Anthropic: https://kasva.devModel: kasva/auto
Dashboard for provider keys. Skip that if you only add credits for our open models.
OpenAI or Anthropic. Same key, same model.
# OpenAI-compatible
curl https://kasva.dev/v1/chat/completions \
-H "Authorization: Bearer YOUR_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"model": "kasva/auto",
"messages": [{"role": "user", "content": "Hello!"}]
}'Free is capped. Pro includes everything in Free, unlimited requests, and priority support.
Our open models: the publisher's list plus 20% for infrastructure (except during sales).