diff --git a/README.md b/README.md
index 3ea406fc..a181d588 100644
--- a/README.md
+++ b/README.md
@@ -378,6 +378,10 @@ default:
 * `git_push_force` – adds `--force-with-lease` to a `git push` (may conflict with `git_push_pull`);
 * `rm_root` – adds `--no-preserve-root` to `rm -rf /` command.
 
+The following rule uses OpenAI ChatGPT. To enable it, you need to set the environment variable `THEFUCK_OPENAI_TOKEN=<OpenAI Token>` and pass in `--chatgpt [No. SUGGESTIONS >= 1]`:
+
+- `chatgpt` &ndash; queries ChatGPT for suggestions. Arguments: `--chatgpt [No. SUGGESTIONS, default=0] --chatgpt-token [default=100] --chatgpt-model [default="gpt-3.5-turbo"]`.
+
 ##### [Back to Contents](#contents)
 
 ## Creating your own rules
diff --git a/requirements.txt b/requirements.txt
index 33ae7a9d..e85ac804 100644
--- a/requirements.txt
+++ b/requirements.txt
@@ -9,3 +9,4 @@ pypandoc
 pytest-benchmark
 pytest-docker-pexpect
 twine
+openai
diff --git a/tests/test_argument_parser.py b/tests/test_argument_parser.py
index e5f1f44b..19fff106 100644
--- a/tests/test_argument_parser.py
+++ b/tests/test_argument_parser.py
@@ -8,7 +8,9 @@ def _args(**override):
             'help': False, 'version': False, 'debug': False,
             'force_command': None, 'repeat': False,
             'enable_experimental_instant_mode': False,
-            'shell_logger': None}
+            'shell_logger': None, 'chatgpt': 0,
+            'chatgpt_token': 100,
+            'chatgpt_model': 'gpt-3.5-turbo'}
     args.update(override)
     return args
 
diff --git a/tests/test_conf.py b/tests/test_conf.py
index e03473ab..98509d60 100644
--- a/tests/test_conf.py
+++ b/tests/test_conf.py
@@ -75,7 +75,7 @@ class TestSettingsFromEnv(object):
 
 
 def test_settings_from_args(settings):
-    settings.init(Mock(yes=True, debug=True, repeat=True))
+    settings.init(Mock(yes=True, debug=True, repeat=True, chatgpt=0, chatgpt_token=100))
     assert not settings.require_confirmation
     assert settings.debug
     assert settings.repeat
diff --git a/thefuck/argument_parser.py b/thefuck/argument_parser.py
index 69c247f1..32cd411c 100644
--- a/thefuck/argument_parser.py
+++ b/thefuck/argument_parser.py
@@ -37,6 +37,24 @@ class Parser(object):
             '-h', '--help',
             action='store_true',
             help='show this help message and exit')
+        self._parser.add_argument(
+            '-c', '--chatgpt',
+            type=int,
+            default=0,
+            help='number of ChatGPT suggestions. set to 0 to disable ChatGPT'
+        )
+        self._parser.add_argument(
+            '-t', '--chatgpt-token',
+            type=int,
+            default=100,
+            help='maximum ChatGPT tokens per query'
+        )
+        self._parser.add_argument(
+            '-m', '--chatgpt-model',
+            type=str,
+            default="gpt-3.5-turbo",
+            help='ChatGPT model'
+        )
         self._add_conflicting_arguments()
         self._parser.add_argument(
             '-d', '--debug',
diff --git a/thefuck/conf.py b/thefuck/conf.py
index 27876ef4..4439147a 100644
--- a/thefuck/conf.py
+++ b/thefuck/conf.py
@@ -124,6 +124,10 @@ class Settings(dict):
             from_args['debug'] = args.debug
         if args.repeat:
             from_args['repeat'] = args.repeat
+
+        from_args['chatgpt'] = args.chatgpt if args.chatgpt >= 0 else 0
+        from_args['chatgpt_token'] = args.chatgpt_token if args.chatgpt_token >= 0 else 0
+        from_args['chatgpt_model'] = args.chatgpt_model
         return from_args
 
 
diff --git a/thefuck/rules/chatgpt.py b/thefuck/rules/chatgpt.py
new file mode 100644
index 00000000..c2161589
--- /dev/null
+++ b/thefuck/rules/chatgpt.py
@@ -0,0 +1,77 @@
+import platform
+import openai
+import re
+import os
+from thefuck import logs
+from thefuck.conf import settings
+
+
+def _check_chatgpt(api_key: str = None) -> bool:
+    openai.api_key = os.getenv("THEFUCK_OPENAI_TOKEN")
+    if settings["chatgpt"] > 0 and (api_key or openai.api_key):
+        return True
+    return False
+
+
+enabled_by_default = _check_chatgpt()
+logs.debug(f"ChatGPT enabled: {enabled_by_default}")
+
+MAX_NUMBER = settings["chatgpt"]
+MAX_TOKENS = settings["chatgpt_token"]
+MODEL = settings["chatgpt_model"]  # by default: "gpt-3.5-turbo"
+
+
+def match(command):
+    return _check_chatgpt()
+
+
+def get_new_command(command):
+    result = _query_chatgpt(
+        command=command.script,
+        error=command.output,
+        explanation=False,
+    )
+    logs.debug(f"chatgpt result: {result}")
+    return result
+
+
+def _query_chatgpt(
+    command: str,
+    error: str,
+    explanation: bool,  # can be used to include explanations but not used yet
+    number: int = MAX_NUMBER,
+    model: str = MODEL,
+    max_tokens: int = MAX_TOKENS,
+    api_key: str = None,
+):
+    if api_key:
+        openai.api_key = api_key
+    elif openai.api_key is None:
+        return []
+
+    os_env = f"{platform.platform()}"
+    prompt = f"""
+OS: `{os_env}`
+Command: `{command}`
+Error: `{error}`
+Suggest {"one command" if number == 1 else f"{number} commands"} {"with" if explanation else "without"} explanation.
+Commands:"""
+
+    logs.debug("chatgpt: " + prompt)
+
+    try:
+        response = openai.ChatCompletion.create(
+            model=model,
+            messages=[
+                {"role": "user", "content": prompt},
+            ],
+            max_tokens=max_tokens,
+        )
+        content = response["choices"][0]["message"]["content"]
+        contents = [item.strip() for item in content.split("\n") if item.strip() != ""]
+        pattern = re.compile(r"^\d+\.\ *")
+        cleaned_contents = [re.sub(pattern, "", item).strip('`') for item in contents]
+        return cleaned_contents
+    except Exception as e:
+        logs.debug(f"chatgpt error: {e}")
+        return []