Troubleshooting SLA Guarantees and API Errors

Understanding SLA Guarantees

Fikra API does not currently publish uptime SLAs, dedicated capacity guarantees, or automatic failover mechanisms. Availability depends on the underlying infrastructure and upstream providers. For production workloads, implement client-side resilience patterns such as retries with exponential backoff.

Common HTTP Error Codes

Exponential Backoff Example (429 and 503)

Use the following Python example to handle rate limits and transient upstream errors:

import time
import random
import requests

FIKRA_API_KEY = "your_api_key_here"  # Load from environment variable in production
BASE_URL = "https://api.fikraapi.co.ke/v1"

def call_fikra_api(payload, max_retries=5):
    url = f"{BASE_URL}/chat/completions"
    headers = {
        "Authorization": f"Bearer {FIKRA_API_KEY}",
        "Content-Type": "application/json"
    }

    for attempt in range(max_retries):
        response = requests.post(url, headers=headers, json=payload)

        if response.status_code in (429, 503):
            # Exponential backoff with jitter
            wait_time = (2 ** attempt) + random.uniform(0, 1)
            print(f"Received {response.status_code}. Retrying in {wait_time:.2f}s...")
            time.sleep(wait_time)
            continue

        if response.status_code == 402:
            raise Exception("Insufficient wallet balance. Please top up in the Fikra dashboard.")

        if response.status_code == 401:
            raise Exception("Invalid or inactive API key. Check your credentials.")

        if response.status_code == 400:
            raise Exception(f"Bad request: {response.text}")

        return response.json()

    raise Exception("Max retries exceeded for 429/503 errors.")

# Example usage
payload = {
    "model": "fikra-flash",
    "messages": [{"role": "user", "content": "Hello"}]
}

try:
    result = call_fikra_api(payload)
    print(result)
except Exception as e:
    print(f"Error: {e}")

Notes

• Rate limits are enforced per account: 30 RPM for unverified accounts and 100 RPM for trusted accounts.
• Do not assume automatic failover or dedicated capacity. Design your application to handle transient failures gracefully.
• For model availability, always check the Fikra model registry in the dashboard.