diff --git a/README.md b/README.md
index c9198ce..fc7ba3b 100644
--- a/README.md
+++ b/README.md
@@ -30,13 +30,15 @@ Explore more options by using `cli-gpt --help`.
    # or (if you have ~/.local/bin created and added to your PATH)
    mv cli-gpt ~/.local/bin
    ```
-5. Configure `cli-gpt` by creating a `~/.cli-gpt` file:
+5. Configure `cli-gpt` by creating a `~/.cli-gpt` file with your desired parameters:
    ```
    OPENAI_API_KEY=<Your API Key>
    MODEL=gpt-4
    ```
    Replace `<Your API Key>` with your API key, and `gpt-4` with `gpt-3.5-turbo` if you don't have access to GPT-4 yet.
 
+   You can also provide additional parameters in the `~/.cli-gpt` file, such as `TEMPERATURE`, `TOP_P`, `N`, `MAX_TOKENS`, `PRESENCE_PENALTY`, `FREQUENCY_PENALTY`, `STOP`, and `LOGIT_BIAS`. To learn more about these parameters, refer to the [OpenAI API documentation](https://platform.openai.com/docs/api-reference/chat/create).
+
 ### Tips
 
 - You can also clone this repo and run `deno install` (follow the instructions on changing your PATH):
diff --git a/src/ChatCompletion.ts b/src/ChatCompletion.ts
index bdcfd6d..06f50de 100644
--- a/src/ChatCompletion.ts
+++ b/src/ChatCompletion.ts
@@ -1,4 +1,5 @@
 import { Message } from './ConversationPersistance.ts';
+import { Config } from './loadConfig.ts';
 
 function normalizeMessages(messages: Message[]): Message[] {
   return messages.reduce<Message[]>((result, message) => {
@@ -13,13 +14,11 @@ function normalizeMessages(messages: Message[]): Message[] {
 }
 
 export class ChatCompletion {
-  private openaiApiKey: string;
-  private model: string;
+  private config: Config;
   private messages: Message[] = [];
 
-  constructor(openaiApiKey: string, model: string = 'gpt-4') {
-    this.openaiApiKey = openaiApiKey;
-    this.model = model;
+  constructor(config: Config) {
+    this.config = config;
   }
 
   setMessages(messages: Message[]) {
@@ -27,15 +26,15 @@ export class ChatCompletion {
   }
 
   async *complete(): AsyncGenerator<string> {
-    const { model, openaiApiKey } = this;
+    const { config: { api_key, ...config } } = this;
     const response = await fetch('https://api.openai.com/v1/chat/completions', {
       method: 'POST',
       headers: {
         'Content-Type': 'application/json',
-        'Authorization': `Bearer ${openaiApiKey}`,
+        'Authorization': `Bearer ${api_key}`,
       },
       body: JSON.stringify({
-        model,
+        ...config,
         messages: normalizeMessages(this.messages),
         stream: true,
       }),
diff --git a/src/index.ts b/src/index.ts
index 1e96075..c43289a 100644
--- a/src/index.ts
+++ b/src/index.ts
@@ -1,23 +1,12 @@
-import { load } from 'std/dotenv/mod.ts';
-import { join } from 'std/path/mod.ts';
 import { writeText as copyToClipboard } from 'copy_paste/mod.ts';
 import { prompt } from './prompt.ts';
 import { ChatCompletion } from './ChatCompletion.ts';
 import { printHelp } from './printHelp.ts';
 import { ConversationPersistance } from './ConversationPersistance.ts';
 import { parseArgs } from './parseArgs.ts';
+import { loadConfig } from './loadConfig.ts';
 
-// TODO: add ability to set the other model params with env vars
-
-const env = await load({
-  envPath: join(Deno.env.get('HOME')!, '.cli-gpt'),
-});
-
-if (env.OPENAI_API_KEY === undefined) {
-  console.error('OPENAI_API_KEY environment variable is not set');
-  Deno.exit(1);
-}
-
+const config = await loadConfig();
 const conversationPersistance = new ConversationPersistance();
 const { flags, role, readFiles, prompt: promptFromArgs } = parseArgs();
 const { affectInitialMessages } = flags;
@@ -55,7 +44,7 @@ if (flags.help) {
   }
 
   if (role === 'user') {
-    const chatCompletion = new ChatCompletion(env.OPENAI_API_KEY, env.MODEL);
+    const chatCompletion = new ChatCompletion(config);
     const encoder = new TextEncoder();
     const write = (chunk: string) => Deno.stdout.write(encoder.encode(chunk));
     const responseContent = [];
diff --git a/src/loadConfig.ts b/src/loadConfig.ts
new file mode 100644
index 0000000..1c2c936
--- /dev/null
+++ b/src/loadConfig.ts
@@ -0,0 +1,108 @@
+import { load } from 'std/dotenv/mod.ts';
+import { join } from 'std/path/mod.ts';
+
+export type Config = {
+  api_key: string;
+  model: string;
+  temperature?: number;
+  top_p?: number;
+  n?: number;
+  max_tokens?: number;
+  presence_penalty?: number;
+  frequency_penalty?: number;
+  stop?: string | string[];
+  logit_bias?: Record<string, number>;
+};
+
+function validateParam<T>(
+  paramName: string,
+  parseValue: () => T | undefined,
+  isValid: (value: T) => boolean,
+): T | undefined {
+  try {
+    const value = parseValue();
+
+    if (value === undefined) {
+      return undefined;
+    }
+
+    if (!isValid(value)) {
+      throw new Error(`Invalid value for parameter ${paramName}: ${value}`);
+    }
+
+    return value;
+  } catch (error) {
+    console.error(`Error parsing parameter ${paramName}: ${error.message}`);
+    Deno.exit(1);
+  }
+}
+
+export async function loadConfig(): Promise<Config> {
+  const env = await load({
+    envPath: join(Deno.env.get('HOME')!, '.cli-gpt'),
+  });
+
+  if (env.OPENAI_API_KEY === undefined) {
+    console.error('OPENAI_API_KEY environment variable is not set');
+    Deno.exit(1);
+  }
+
+  return {
+    api_key: env.OPENAI_API_KEY,
+    model: env.MODEL ?? 'gpt-4',
+    temperature: validateParam<number>(
+      'temperature',
+      () =>
+        env.TEMPERATURE !== undefined ? parseFloat(env.TEMPERATURE) : undefined,
+      (value) => value >= 0 && value <= 2,
+    ),
+    top_p: validateParam<number>(
+      'top_p',
+      () => (env.TOP_P !== undefined ? parseFloat(env.TOP_P) : undefined),
+      (value) => value >= 0 && value <= 1,
+    ),
+    n: validateParam<number>(
+      'n',
+      () => (env.N !== undefined ? parseInt(env.N) : undefined),
+      (value) => value > 0,
+    ),
+    max_tokens: validateParam<number>(
+      'max_tokens',
+      () => (env.MAX_TOKENS !== undefined ? parseInt(env.MAX_TOKENS) : undefined),
+      (value) => value >= 0,
+    ),
+    presence_penalty: validateParam<number>(
+      'presence_penalty',
+      () =>
+        env.PRESENCE_PENALTY !== undefined
+          ? parseFloat(env.PRESENCE_PENALTY)
+          : undefined,
+      (value) => value >= -2 && value <= 2,
+    ),
+    frequency_penalty: validateParam<number>(
+      'frequency_penalty',
+      () =>
+        env.FREQUENCY_PENALTY !== undefined
+          ? parseFloat(env.FREQUENCY_PENALTY)
+          : undefined,
+      (value) => value >= -2 && value <= 2,
+    ),
+    stop: validateParam<string | string[]>(
+      'stop',
+      () => (env.STOP !== undefined ? JSON.parse(env.STOP) : undefined),
+      (value) =>
+        (typeof value === 'string' && value.trim() !== '') ||
+        (Array.isArray(value) && value.length > 0 && value.length <= 4),
+    ),
+    logit_bias: validateParam<Record<string, number>>(
+      'logit_bias',
+      () => (env.LOGIT_BIAS !== undefined ? JSON.parse(env.LOGIT_BIAS) : undefined),
+      (value) => {
+        for (const bias of Object.values(value)) {
+          if (bias < -100 || bias > 100) return false;
+        }
+        return true;
+      },
+    ),
+  }
+}