Baseten Alternative: Fikra API Integration Guide

Introduction

Baseten is an external inference platform that hosts and serves various machine learning models. Developers often look for alternatives when seeking different integration patterns, regional considerations, or specific model availability. Fikra API is a separate service that provides access to its own set of models (such as fikra-flash and fikra-pro-20b) through an OpenAI-compatible interface. This guide explains how to integrate with Fikra API if you are evaluating options beyond Baseten.

Comparison: Baseten vs. Fikra API

Note: Fikra API does not host, resell, or provide Baseten’s models. The comparison above focuses on integration and access patterns.

Python Example: Using Fikra API with OpenAI SDK

Fikra API supports OpenAI-style requests. You can use the standard openai Python SDK by pointing the base URL to Fikra’s endpoint.

import os
from openai import OpenAI

# Initialize the client with Fikra's base URL and API key
client = OpenAI(
    base_url="https://api.fikraapi.co.ke/v1",
    api_key=os.getenv("FIKRA_API_KEY")
)

# Make a chat completion request using a Fikra model alias
response = client.chat.completions.create(
    model="fikra-flash",
    messages=[
        {"role": "system", "content": "You are a helpful assistant."},
        {"role": "user", "content": "Hello, Fikra!"}
    ]
)

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

Streaming Support

Fikra API supports streaming via Server-Sent Events (SSE). Set stream=True in your request to receive incremental responses.

stream = client.chat.completions.create(
    model="fikra-flash",
    messages=[{"role": "user", "content": "Tell me a short story."}],
    stream=True
)

for chunk in stream:
    if chunk.choices[0].delta.content:
        print(chunk.choices[0].delta.content, end="")

When Fikra API is a Fit

Consider Fikra API if: - You prefer an OpenAI-compatible interface for easier migration from existing OpenAI-based code. - You are based in or serving the Kenyan market and want local payment options (M-Pesa and cards via Paystack). - You need access to Fikra’s specific model aliases (e.g., fikra-flash for speed, fikra-pro-120b for higher capacity). - You want a token-credit wallet system for managing usage costs.

Verify Before Production

Before deploying Fikra API in production, please: 1. Check the Model Registry: Confirm that the specific Fikra model alias you need is currently available and enabled for your account. 2. Review Rate Limits: Current public limits are 30 RPM for unverified accounts and 100 RPM for trusted accounts. Ensure your application’s traffic patterns fit within these limits. 3. Test Authentication: Verify that your FIKRA_API_KEY is correctly configured and has sufficient token credits. 4. Monitor Usage: Use the Fikra dashboard to track token consumption and wallet balance. 5. Consult Documentation: Refer to the official Fikra API documentation at https://fikraapi.co.ke for the latest endpoint details and error codes.

Fikra API does not guarantee specific uptime, SLAs, or performance metrics. Always test in a staging environment before production deployment.