Integrating Fikra API with LlamaIndex

This guide demonstrates how to use the Fikra API as a custom OpenAI-compatible endpoint within the LlamaIndex framework. By configuring the base_url, you can leverage Fikra's models for your RAG and agent applications.

Prerequisites

1. Install the required packages: bash pip install llama-index-llms-openai
2. Obtain your Fikra API key from your dashboard.

Step 1: Import Dependencies

Import the necessary classes from LlamaIndex and OpenAI.

from llama_index.llms.openai import OpenAI

Step 2: Configure the Custom Endpoint

Instantiate the OpenAI LLM class. The key step is setting the base_url parameter to point to the Fikra API endpoint: https://api.fikraapi.co.ke/v1.

llm = OpenAI(
    model="gpt-4o-mini", # Or any supported model
    api_key="your_fikra_api_key",
    base_url="https://api.fikraapi.co.ke/v1"
)

Step 3: Use the LLM

You can now use the configured LLM instance in your LlamaIndex workflows.

response = llm.complete("Hello, Fikra! How are you?")
print(response.text)

JavaScript/TypeScript Example

If you are using the LlamaIndex JavaScript/TypeScript SDK, the configuration is similar.

import { OpenAI } from "@llamaindex/llms-openai";

const llm = new OpenAI({
  model: "gpt-4o-mini",
  apiKey: "your_fikra_api_key",
  baseURL: "https://api.fikraapi.co.ke/v1"
});

const response = await llm.complete("Hello, Fikra!");
console.log(response.text);

Notes

• Ensure your API key has sufficient credits.
• The base_url must include the /v1 suffix as shown above.
• Refer to the Fikra API Documentation for available models and rate limits.