Overview

Fikra API is an OpenAI-compatible AI inference API for developers and growing AI teams who need production inference without a credit-card gate. It exposes chat and embedding models behind one OpenAI-style endpoint, and billing happens through a token-credit wallet in USD or KES. To stay on a fixed monthly cap, pick a model with known per-token pricing, send requests through the standard OpenAI SDK, and monitor token usage from each response.

Prerequisites

A Fikra API key, stored in the environment variable FIKRA_API_KEY
Python 3.8+ with the openai SDK installed
A token-credit wallet with funds (top-ups are available via M-Pesa and card through Paystack, or via x402 USDC on Base for agents)

Step-by-step

Configure the OpenAI client to point at Fikra API:

import os
from openai import OpenAI

client = OpenAI(
    base_url="https://api.fikraapi.co.ke/v1",
    api_key=os.environ["FIKRA_API_KEY"],
)

Choose a model from the pricing list. For a low-cost budget, deepseek-v4-flash-0731 costs $0.09 per 1M input tokens and $0.15 per 1M output tokens.
Send a chat completion request:

response = client.chat.completions.create(
    model="deepseek-v4-flash-0731",
    messages=[
        {"role": "user", "content": "Summarize the budget policy in one sentence."}
    ],
)

Extract the reply and usage from the response, then calculate the cost of the call:

print(response.choices[0].message.content)

usage = response.usage
input_cost = usage.prompt_tokens / 1_000_000 * 0.09
output_cost = usage.completion_tokens / 1_000_000 * 0.15
print(f"Cost: ${input_cost + output_cost:.4f}")

Track each call's cost against your monthly cap. Because the API response includes prompt_tokens and completion_tokens, you can accumulate these amounts client-side and stop or alert when the cap is reached.
Control spend at the wallet level by topping up only the amount allocated for the month. Fikra API operates on a token-credit wallet, so a fixed top-up acts as your spending ceiling.

Verify it works

Run the script. You should see a plain-text reply from the model, followed by the calculated cost. The request and response are OpenAI-style chat completions, so the same pattern works across tools that accept an OpenAI-compatible base URL, including LangChain, Vercel AI SDK, OpenAI SDK, Cursor, and CrewAI.

Costs

Pricing is per 1M tokens. Current model rates:

Model	Input (USD)	Output (USD)
deepseek-v4-flash-0731	0.09	0.15
deepseek-v4.1-flash	0.25	0.80
deepseek-v4.1-flash-1017	0.30	0.85
fikra-auto	0.40	0.40
fikra-auto-pro	6.00	12.00
fikra-code-35b	1.00	1.00

For any model or rate not listed here, check the pricing page.

Common errors

401 Unauthorized to The API key is missing, invalid, or inactive. Confirm that FIKRA_API_KEY is set and that the key is scoped and revocable.
400 Bad Request to The model name is not supported. Use one of the model names from the pricing table above.
429 Too Many Requests to You hit the rate limit: 30 requests/min for unverified keys, 100 requests/min for trusted keys. Slow down or request trusted status.
503 Service Unavailable to The upstream is busy. Retry with backoff.

How to Run an OpenAI-Compatible Budget on a Fixed Monthly Cap

Fact block, plain text for agents