Pricing
How SoxAI charges per request, how to see the exact cost, and how the launch-promo discount is applied.
Pricing
SoxAI charges per-token based on the upstream model you use. There is no plan tier and no per-seat fee — one account, one balance, per-request billing.
How the price is built
Your cost = tokens × model rate × (1 − launch promo discount)- Model rate — every model in the catalog has a published input rate and output rate (USD per 1M tokens). Same across all accounts; no hidden markup.
- Launch promo discount — 15% off every model during the current launch promo. Applied at settlement, not at quote time. Ends on a published schedule with at least 30 days of notice.
Example calculation
For gpt-5.4-mini (list rate: $0.75 input, $4.50 output per 1M tokens) with the 15% launch promo:
Input: 1,000 tokens × $0.75/1M × 0.85 = $0.0006375
Output: 500 tokens × $4.50/1M × 0.85 = $0.0019125
Total request cost = $0.002550The 15% discount applies automatically; it shows up as a line item on your invoice, not as a separate rate.
Where to see the exact cost per request
Console → My Logs shows every request with its computed cost in USD. Useful for:
- Spotting runaway loops or mis-routed traffic
- Debugging "why did this one request cost $X"
- Exporting CSV for finance reconciliation
If you're integrating deeply, the same data is available through the billing transaction API — see the Invoices guide for the endpoint.
Free tier
- $0.50 trial credit on signup. No credit card required.
- $5 bonus on your first top-up of $5 or more.
The trial credit is enough to exercise the full flow (create a key, make requests, see them in the logs) on most models. The first-topup bonus is how we compensate for not requiring a card upfront.
Pricing transparency
- Live catalog (JSON) — authoritative model IDs, context windows, and per-million input/output rates, updated in real time.
- Supported models — browseable snapshot grouped by provider.
- /models — marketing-friendly model browser with search.
If you see a price that doesn't match your invoice, email [email protected] with the request ID from My Logs and we'll reconcile.