API pricing

Pay per token. Nothing you did not spend.

Prepaid credit, drawn down per token. $5 to start with no card, published rate limits, and no overage bill — when the balance runs out, requests stop.

tase-0.1-8b

Tool calling, low latency, Turkish + English

128K context

Input

$0.15

per 1M tokens

Output

$0.60

per 1M tokens

A typical assistant call — 800 input tokens, 200 generated — costs about $0.00024. That is roughly 4,167 calls per dollar.

Credit packs

Buy credit when you need it. One-time payment, no subscription, no expiry. Every account starts with $5 already on it.

$10

67M input tokens, or 41,667K typical calls

$25

167M input tokens, or 104,167K typical calls

Unlocks Scale limits

$100

667M input tokens, or 416,667K typical calls

Unlocks Scale limits

$500

3,333M input tokens, or 2,083,333K typical calls

Unlocks Scale limits

Rate limits by plan

These are the numbers the limiter actually enforces, not a guideline. Scale applies automatically at $20 in lifetime top-ups.

PlanRequests / minConcurrencyKeysAgents
Free20235
Scale120122050
EnterpriseNegotiatedNegotiated100250

Included on every plan

Chat completions, streaming, and tool calling
Stored agents with their own run history
Request logs with tokens, cost, and latency
Usage API for your own dashboards
Rate-limit headers on every response
No charge for failed requests

Questions

How is a request priced?+

Per token, split into input and output. A request that sends 800 input tokens and generates 200 output tokens costs 800 × $0.15/1M + 200 × $0.60/1M — about $0.00024. The exact charge, per request, is in your console logs.

What is the $5 starter credit?+

Every account is granted $5 the first time it opens the console. No card, no trial clock, and it does not expire. At list price that is roughly 33 million input tokens.

Can I get an overage bill?+

No. Credit is prepaid, so when the balance reaches zero the API returns 402 and stops. That is the whole billing model — there is no invoice arriving later for traffic you did not expect.

Does credit expire?+

No. Credit sits on the account until it is spent.

How do I get higher rate limits?+

Scale unlocks automatically once lifetime top-ups reach $20 — 120 requests per minute per key and 12 concurrent requests. Above that, Enterprise limits are negotiated with invoiced billing instead of prepaid credit.

Do workbench runs cost money?+

Yes, and deliberately so. The console workbench uses the same endpoint, model, and meter as production, so what you measure there is what you will pay in production.

Start on the house.

$5 of credit lands the first time you open the console. Spend it before you decide.

Open the console