How does billing and token tracking work?
Fikra API uses a predictable pay-as-you-go billing model. Your account maintains a credit balance that can be funded through supported payment methods. API usage consumes credits according to the pricing of the model you use.
Token Economics
Fikra API uses model-specific token pricing rather than a single universal rate. Different models have different computational characteristics and therefore different prices. The current authoritative pricing for available models is published in the Model Registry.
| Pricing Tier | Current Models | Price per 1M Tokens (USD) |
|---|---|---|
| Live | deepseek-v4.1-flash: | $ 0.05 in / 0.15 out |
| Live | deepseek-v4.1-flash-1017: The latest DeepSeek V4.1 Flash update (Oct 17 build): better reasoning | $ 0.05 in / 0.15 out |
| Live | deepseek-v4-flash-0731: | $ 0.05625 in / 0.17 out |
| Live | glm-5.3-flash: | $ 0.155 in / 0.49 out |
| Live | kimi-k3: | $ 2.1 in / 10.5 out |
| Live | step-5-preview: | $ 1.25 in / 3.3 out |
| Live | mimo-v2.6-pro: Xiaomi MiMo-V2.6-Pro: the top-scoring open-weight model on the Artific | $ 0.5 in / 1.0 out |
| Live | deepseek-v4-flash-ark: DeepSeek V4 Flash served through BytePlus ModelArk. 1M context, 393k o | $ 0.55 in / 1.65 out |
| Live | gemini-4-argon: | $ 2.5 in / 12.0 out |
| Live | glm-5.3: Z.ai flagship GLM-5.3: large-scale reasoning for complex software engi | $ 1.05 in / 3.35 out |
| Live | glm-5.2: Z.ai GLM-5.2 on a 1M context: strong coding and reasoning at a lower p | $ 0.75 in / 1.65 out |
| Live | nvidia-nemotron-3.5-super-vl: | $ 0.5 in / 1.25 out |
| Live | qwen3.8-27b: Fast 27B-class multimodal model with automatic prompt caching. Cached | $ 0.045 in / 0.35 out |
| Live | deepseek-v4-pro-ark: DeepSeek V4 Pro served through BytePlus ModelArk. 1M context, 393k out | $ 1.65 in / 4.95 out |
| Live | deepseek-v4.1-flash-ark: DeepSeek V4.1 Flash served through BytePlus ModelArk. 1M context, 393k | $ 0.1875 in / 0.75 out |
| Live | deepseek-v4-1-flash-economy-free: Temporary free promotional route backed by BytePlus ModelArk. Availabl | $ 0.0 in / 0.0 out |
| Live | fikra-pro-20b: | $ 0.2 in / 0.2 out |
| Live | fikra-pro-120b: | $ 0.3 in / 0.3 out |
| Live | fikra-qwen-27b: | $ 0.045 in / 0.35 out |
| Live | fikra-qwen-max: | $ 6.0 in / 12.0 out |
| Live | fikra-auto: | $ 0.045 in / 0.35 out |
| Live | fikra-auto-pro: | $ 6.0 in / 12.0 out |
| Live | fikra-code-35b: | $ 0.045 in / 0.33 out |
Fikra API pricing can change as models and infrastructure evolve. The Model Registry is the authoritative reference for current model availability and pricing. Check the registry before making production cost assumptions.
Account Credits & Usage
Fikra API separates payment funding from model usage. You add funds to your account and those funds become available as account credits. As requests are processed, the corresponding usage cost is deducted from your available balance.
The Billing Lifecycle
- Add funds to your Fikra API account using a supported payment method.
- Your account receives the corresponding billing credit.
- API requests consume credits according to the selected model's published pricing.
- Your remaining balance is reflected in your account and billing information.
- If your available balance is exhausted, API requests requiring additional paid usage will be rejected until the account is funded again.
Billing is separate from model selection
You can use different Fikra models from the same account. The amount consumed depends on the pricing associated with the model handling the request. Switching models therefore changes the cost of subsequent usage, while your account and billing balance remain the same.
Payments
Fikra API uses Paystack to process account payments. This allows customers to fund their accounts using locally relevant payment methods as well as other payment methods supported by Paystack.
Supported Payment Infrastructure
- 01 Paystack handles payment processing and transaction confirmation.
- 02 Card and mobile money Both are charged in KES at KES 133 = $1 through Paystack, and the wallet is credited in USD. New accounts are prepaid; existing subscriptions are honored.
- 03 Additional payment methods may be introduced as Fikra expands its payment infrastructure.
Payment processing is handled securely through the payment infrastructure provided by Paystack. Fikra does not require customers to expose payment credentials to the Fikra inference API itself.
Adding Funds
Account funding is performed through the Fikra billing experience. When a payment is successfully completed, the corresponding amount is applied to the customer's account balance.
Typical Top-Up Flow
- Open the billing section of your Developer Dashboard .
- Select the amount you want to add to your account.
- Choose a supported payment method and complete the payment through the secure payment interface.
- Once the transaction is confirmed, the funds are reflected in your Fikra account.
- You can then continue making API requests against your available balance.
The minimum account funding amount is $1 USD (KES 133), charged in KES.
Fikra API is fully self-serve: a prepaid wallet, pay-as-you-go token pricing and rate limits that scale automatically on your first top-up. New accounts are prepaid. Existing subscriptions are honored, and API usage is billed according to live model pricing.
| Plan | Monthly Platform Fee | Capacity |
|---|---|---|
| Self-serve | $0 / month | Pay-as-you-go prepaid wallet. Trusted-tier limits unlock on your first top-up. |
Platform fees are not token credits
Wallet top-ups are consumed strictly per usage: every request is metered against the model's published rate and deducted from your prepaid balance in real time.
Automatic Recharge
Fikra API can support automatic account funding for eligible payment methods. When enabled, automatic recharge can help maintain service continuity when an account approaches its configured balance threshold.
Availability
Automatic recharge availability depends on the payment method and account configuration. Payment credentials are handled through the payment provider, rather than being exposed to the Fikra inference API.
When Your Balance Runs Out
Paid API requests require sufficient account balance. When your available balance
is insufficient for continued usage, Fikra API returns
402 Payment Required.
Payment Required
Your Fikra account does not have sufficient available balance to process the requested paid usage. Add funds to your account and retry the request.
Pricing Guidance
Model pricing is maintained independently from this billing overview because Fikra's model catalogue evolves over time. For production applications, always use the Model Registry as the current reference for model names, availability, and pricing.
Keep these concepts separate
- • Payment is how you add funds to your account.
- • Account balance represents the funds available for paid usage.
- • Token pricing determines how much model usage consumes from your balance.
- • Platform plans such as Growth and Scale provide additional capacity and service commitments separately from inference usage.