Pay per token. The first 3M a day are free.
No subscription, no minimum, no tiers. Rates are per 1M tokens and the same for every key.
Free every day
- 3M tokens per account per day, reset at 00:00 UTC
- Free models never draw on the pool
- No card to start
- Paid models draw on prepaid credit only after the pool is spent
How you pay
Prepaid credit in USD, topped up from the Billing page. Credit never expires.
Get an API key| Model | Context | Input / 1M | Output / 1M |
|---|---|---|---|
| aetron/qwen3.5-9b-gguf-q4-k-m Qwen3.5-9B (GGUF Q4_K_M) | 256K | free | |
| aetron/qwen3.8-27b-gguf-q4-k-m Qwen3.8-27B (GGUF UD-Q4_K_M) | 256K | free | |
Auto-router: picks a connected model for each request; you pay the rate of the model it picked.
Pick a preset or enter exact token counts. The estimate uses the rates of the model you choose.
What counts toward the free pool?
Input and output tokens of paid models. Traffic to free models is not counted.
What happens when credit runs out?
The daily free pool and free models keep working. Paid requests beyond the pool return 402 insufficient_credits until you top up.
Is aetron/auto free?
The slug has no rate of its own. You pay the rate of the model the router picked; it shows in the request log.
Can I cap what a key can spend?
Yes: tokens, requests and USD per day per key, plus an IP allowlist. Set them on the API keys page.
Is there a subscription?
No. Pay-as-you-go only, with the same rates for everyone.