diff --git a/llama_stack/cli/info.py b/llama_stack/cli/info.py
new file mode 100644
index 0000000000..2ba545e8f9
--- /dev/null
+++ b/llama_stack/cli/info.py
@@ -0,0 +1,114 @@
+# Copyright (c) Meta Platforms, Inc. and affiliates.
+# All rights reserved.
+#
+# This source code is licensed under the terms described in the LICENSE file in
+# the root directory of this source tree.
+
+import argparse
+import importlib
+import platform
+import subprocess
+import sys
+
+import psutil
+
+from llama_stack.cli.subcommand import Subcommand
+
+
+def _system_info():
+    sys_info = {
+        "sys.version": sys.version,
+        "sys.platform": sys.platform,
+        "platform.machine": platform.machine(),
+        "platform.node": platform.node(),
+        "platform.python_version": platform.python_version(),
+    }
+    if sys.platform == "linux":
+        os_release = platform.freedesktop_os_release()
+        for key in ["ID", "VERSION_ID", "PRETTY_NAME"]:
+            value = os_release.get(key)
+            if value:
+                sys_info[f"os-release.{key}"] = value
+    elif sys.platform == "darwin":
+        try:
+            cpu_info = subprocess.check_output(["sysctl", "-n", "machdep.cpu.brand_string"]).decode("utf-8").strip()
+            sys_info["platform.cpu_brand"] = cpu_info
+        except subprocess.CalledProcessError:
+            sys_info["platform.cpu_brand"] = "Unknown"
+    sys_memory_info = psutil.virtual_memory()
+    sys_info["memory.used"] = f"{(sys_memory_info.used / 1024**3):.2f} GB"
+    sys_info["memory.available"] = f"{(sys_memory_info.available / 1024**3):.2f} GB"
+    sys_info["memory.total"] = f"{(sys_memory_info.total / 1024**3):.2f} GB"
+    return sys_info
+
+
+def _list_pkg_version(pkg_name):
+    packages = sorted(
+        (distribution.name, distribution.version)
+        for distribution in importlib.metadata.distributions()
+        if distribution.name.startswith(pkg_name)
+    )
+    return {f"{name}.version": version for name, version in packages}
+
+
+def _cuda_info():
+    import torch
+
+    if torch.cuda.is_available():
+        cuda_info = {
+            "cuda.version": torch.version.cuda,
+            "gpu.count": torch.cuda.device_count(),
+            "cuda.bf16": torch.cuda.is_bf16_supported(),
+            "cuda.current.device": torch.cuda.current_device(),
+        }
+        for idx in range(torch.cuda.device_count()):
+            device = torch.device("cuda", idx)
+            free, total = torch.cuda.mem_get_info(device)
+            cuda_info[f"cuda.{idx}.name"] = torch.cuda.get_device_name(device)
+            cuda_info[f"cuda.{idx}.free"] = f"{(free / 1024**3):.1f} GB"
+            cuda_info[f"cuda.{idx}.total"] = f"{(total / 1024**3):.1f} GB"
+        return cuda_info
+
+
+def _add_to_group(groups, item_name, info):
+    if item_name not in groups:
+        groups[item_name] = []
+    groups[item_name].extend(info.items())
+
+
+def _get_env_info_by_group():
+    groups = {}
+    _add_to_group(groups, "system", _system_info())
+    _add_to_group(groups, "llama packages", _list_pkg_version("llama"))
+    _add_to_group(groups, "torch packages", _list_pkg_version("torch"))
+
+    # only cuda exist
+    cuda_info = _cuda_info()
+    if cuda_info:
+        _add_to_group(groups, "cuda info", _cuda_info())
+
+    return groups
+
+
+class LlamaStackInfo(Subcommand):
+    """Llama cli for showing llama stack environment information"""
+
+    def __init__(self, subparsers: argparse._SubParsersAction):
+        super().__init__()
+        self.parser = subparsers.add_parser(
+            "info",
+            prog="llama info",
+            description="Show llama stack environment information",
+            formatter_class=argparse.RawDescriptionHelpFormatter,
+        )
+        self._add_arguments()
+        self.parser.set_defaults(func=self._run_llama_stack_info_cmd)
+
+    def _run_llama_stack_info_cmd(self, args: argparse.Namespace) -> None:
+        groups = _get_env_info_by_group()
+        for idx, (group, items) in enumerate(groups.items()):
+            if idx > 0:
+                print()
+            print(f"{group}:")
+            for k, v in items:
+                print(f"  {k}: {v}")
diff --git a/llama_stack/cli/llama.py b/llama_stack/cli/llama.py
index 8ff5800298..0890b6ac78 100644
--- a/llama_stack/cli/llama.py
+++ b/llama_stack/cli/llama.py
@@ -7,6 +7,7 @@
 import argparse
 
 from .download import Download
+from .info import LlamaStackInfo
 from .model import ModelParser
 from .stack import StackParser
 from .stack.utils import print_subcommand_description
@@ -34,6 +35,7 @@ def __init__(self):
         StackParser.create(subparsers)
         Download.create(subparsers)
         VerifyDownload.create(subparsers)
+        LlamaStackInfo.create(subparsers)
 
         print_subcommand_description(self.parser, subparsers)