Billing
How token cost is calculated, the 1.09 margin, reasoning-effort multipliers, viewing spend at /cabinet/logs, and top-up methods.
Payment model
Codex Key is prepaid: you top up an internal RUB balance, and every request automatically deducts its token cost. There are no subscriptions, no minimum payments, no required tariffs.
Cost formula
The cost of a single request is computed as:
cost_rub = (
prompt_tokens * price_in_per_token
+ output_tokens * price_out_per_token * effort_multiplier
+ reasoning_tokens * price_out_per_token
) * margin * usd_to_rub
Where:
price_in_per_token,price_out_per_token— OpenAI base prices per token (USD).effort_multiplier— multiplier fromreasoning_effort(see below).margin= 1.09 (fixed platform margin).usd_to_rub— CBR exchange rate at request time.
All three usage components (prompt / completion / reasoning) are billed. Reasoning tokens are billed as output.
Model coefficients
| Model | Base coefficient |
|---|---|
gpt-5.4 | 1.0× |
gpt-5.5 | 1.4× |
gpt-5.4-mini | 0.2× |
codex-5.3 | 1.1× |
The coefficient is applied to the OpenAI price before margin. See the models reference for details.
reasoning_effort multipliers
| Effort | Output multiplier |
|---|---|
minimal | 1.0× |
low | 1.5× |
medium | 2.5× |
high | 4.0× |
With effort = high a single completion can cost 4× more than with minimal. Use it deliberately.
Service tier
| Tier | Total multiplier |
|---|---|
default | 1.0× |
priority | 1.3× |
flex | 0.6× |
Calculation example
A request to gpt-5.4 with reasoning_effort = medium:
- prompt: 1000 tokens × $0.000002 = $0.002
- output: 500 tokens × $0.000010 × 2.5 = $0.0125
- subtotal: $0.0145
- × margin 1.09 = $0.01580
- × FX rate 100 RUB/USD = 1.58 RUB
Actual prices and FX rates fluctuate; the real cost of each request is shown in /cabinet/logs.
Viewing spend
Several dashboard sections track usage:
/cabinet— current balance and spend for today / 7 days / 30 days./cabinet/logs— every request, filterable by date, model, key, status./cabinet/usage— charts by model and key./cabinet/limits— current RPM/TPM limits and monthly spend.
Each log entry shows:
- timestamp, model, key (last 4 characters),
- prompt / completion / reasoning / total tokens,
- cost in RUB,
- HTTP status and duration.
Logs can be exported as CSV for any period.
Top-up methods
The /cabinet/billing section supports:
- Bank card (YooKassa, RU/Visa/Mastercard) — instant credit.
- SBP (Russian instant payments) — instant credit via QR.
- USDT (TRC-20) — credited after 1 network confirmation (~1 min).
- Bank transfer for legal entities — invoice + reconciliation, 1–3 business days.
Minimum amount: 100 RUB. Maximum single operation: 500,000 RUB (higher on request to support).
Refunds
Refunds of unused balance are available within 14 days of top-up by request to support@codexkey.ru. After 14 days or with active usage (>50% spent) refunds are not issued. See the Offer and Refund policy for details.
Alerts and limits
In /cabinet/billing you can configure:
- email alert when balance falls below X RUB,
- automatic disabling of all keys when balance hits 0,
- soft daily spend cap (requests return 429 after the cap is hit).