src/pages/api/llm.js

import { OpenAIError, OpenAIStream } from '../../stream';
import wasm from '../../../node_modules/@dqbd/tiktoken/lite/tiktoken_bg.wasm?module';
import tiktokenModel from '@dqbd/tiktoken/encoders/cl100k_base.json';
import { Tiktoken, init } from '@dqbd/tiktoken/lite/init';

export const DEFAULT_SYSTEM_PROMPT =
  process.env.NEXT_PUBLIC_DEFAULT_SYSTEM_PROMPT ||
  "You are ChatGPT, a large language model trained by OpenAI. Follow the user's instructions carefully. Respond using markdown.";

export const OPENAI_API_HOST =
  process.env.OPENAI_API_HOST || 'https://api.openai.com';

export const DEFAULT_TEMPERATURE = 
  parseFloat(process.env.NEXT_PUBLIC_DEFAULT_TEMPERATURE || "1");

export const OPENAI_API_TYPE =
  process.env.OPENAI_API_TYPE || 'openai';

export const OPENAI_API_VERSION =
  process.env.OPENAI_API_VERSION || '2023-03-15-preview';

export const OPENAI_ORGANIZATION =
  process.env.OPENAI_ORGANIZATION || '';

export const AZURE_DEPLOYMENT_ID =
  process.env.AZURE_DEPLOYMENT_ID || '';

export const config = {
  runtime: 'edge',
};

const handler = async (req) => {
  try {
    const { model, messages, key, prompt, temperature } = (await req.json());

    await init((imports) => WebAssembly.instantiate(wasm, imports));
    const encoding = new Tiktoken(
      tiktokenModel.bpe_ranks,
      tiktokenModel.special_tokens,
      tiktokenModel.pat_str,
    );

    let promptToSend = prompt;
    if (!promptToSend) {
      promptToSend = DEFAULT_SYSTEM_PROMPT;
    }

    let temperatureToUse = temperature;
    if (temperatureToUse == null) {
      temperatureToUse = DEFAULT_TEMPERATURE;
    }

    const prompt_tokens = encoding.encode(promptToSend);

    let tokenCount = prompt_tokens.length;
    let messagesToSend = [];

    for (let i = messages.length - 1; i >= 0; i--) {
      const message = messages[i];
      const tokens = encoding.encode(message.content);

      if (tokenCount + tokens.length + 1000 > model.tokenLimit) {
        break;
      }
      tokenCount += tokens.length;
      messagesToSend = [message, ...messagesToSend];
    }

    encoding.free();

    const stream = await OpenAIStream(model, promptToSend, temperatureToUse, key, messagesToSend);

    return new Response(stream);
  } catch (error) {
    console.error(error);
    if (error instanceof OpenAIError) {
      return new Response('Error', { status: 500, statusText: error.message });
    } else {
      return new Response('Error', { status: 500 });
    }
  }
};

export default handler;