diff --git a/neural-networks/generic-example/AGENTS.md b/neural-networks/generic-example/AGENTS.md
index af1af7792..dfac644eb 100644
--- a/neural-networks/generic-example/AGENTS.md
+++ b/neural-networks/generic-example/AGENTS.md
@@ -25,7 +25,8 @@ Reusable single-model inference scaffold. It runs one Model Zoo model with one i
 - `Default RVC2 model:` [depthai_models/yolov6_nano_r2_coco.RVC2.yaml](depthai_models/yolov6_nano_r2_coco.RVC2.yaml)
 - `Default RVC4 model:` [depthai_models/yolov6_nano_r2_coco.RVC4.yaml](depthai_models/yolov6_nano_r2_coco.RVC4.yaml)
 - `Standalone config:` [oakapp.toml](oakapp.toml)
-- `Default model slug:` `luxonis/yolov6-nano:r2-coco-512x288`
+- `Default model identifier:` `luxonis/yolov6-nano:r2-coco-512x288`
+- `Accepted model inputs:` Model Zoo/HubAI model identifier, local model descriptor/YAML, or local `.tar.xz` `NNArchive`
 - `Input:` camera by default, or media file via `--media_path`
 - `Output:` `Video` passthrough plus parsed output on `Detections`
 
@@ -33,13 +34,14 @@ Reusable single-model inference scaffold. It runs one Model Zoo model with one i
 
 - [main.py](main.py): model selection, pipeline construction, parser node, and Visualizer topics
 - [utils/input.py](utils/input.py): camera versus `ReplayVideo` input selection and platform-specific frame type
-- [utils/arguments.py](utils/arguments.py): model slug, FPS limit, media path, API key, and overlay options
+- [utils/arguments.py](utils/arguments.py): model source, FPS limit, media path, API key, and overlay options
 - [oakapp.toml](oakapp.toml): standalone entrypoint and packaged defaults
 
 ## Architecture
 
 - [main.py](main.py) connects to a device, reads the platform string, and chooses `yolov6_nano_r2_coco.<platform>.yaml` as the default descriptor.
-- If `--model` differs from the YAML-backed default, [main.py](main.py) creates `dai.NNModelDescription(args.model, platform=platform)` instead.
+- If `--model` ends with `.tar.xz`, [main.py](main.py) loads it directly as an `NNArchive`.
+- Otherwise, if `--model` differs from the YAML-backed default model identifier, [main.py](main.py) creates `dai.NNModelDescription(args.model, platform=platform)` and resolves it through the Model Zoo.
 - [utils/input.py](utils/input.py) returns either a camera node or a `ReplayVideo` node.
 - `ParsingNeuralNetwork` runs the model and emits parsed output.
 - `Video` shows `ParsingNeuralNetwork.passthrough` unless `--overlay_mode` is enabled.
@@ -55,13 +57,13 @@ Reusable single-model inference scaffold. It runs one Model Zoo model with one i
 
 ## Modification Guide
 
-- `Safe to change:` model slug, FPS limit, media path, private API key handling, Visualizer topic names
+- `Safe to change:` model source, FPS limit, media path, private API key handling, Visualizer topic names
 - `Requires care:` model input/output compatibility, overlay assumptions, platform-specific media frame types, standalone defaults in [oakapp.toml](oakapp.toml)
 - `Likely to break if changed blindly:` using multi-input or multi-head models, assuming every parsed output is a detection, enabling overlay for non-image-like outputs
 
 ## Common Adaptations
 
-- `Swap the model:` pass `--model` first; edit YAMLs only if changing the default packaged baseline.
+- `Swap the model:` pass `--model` first; this can be a Model Zoo/HubAI model identifier, a local YAML/descriptor, or a local `.tar.xz` archive. Edit YAMLs only if changing the default packaged baseline.
 - `Run on media:` pass `--media_path`; [utils/input.py](utils/input.py) switches from `Camera` to `ReplayVideo`.
 - `Use a private model:` set `--api_key` or `DEPTHAI_HUB_API_KEY`.
 - `Create a task-specific example:` keep [main.py](main.py), [utils/input.py](utils/input.py), and [utils/arguments.py](utils/arguments.py), then replace parser/output handling as needed.
@@ -85,5 +87,6 @@ Reusable single-model inference scaffold. It runs one Model Zoo model with one i
 
 - `Run:` `python3 main.py`
 - `Alternative run:` `python3 main.py --model luxonis/mediapipe-selfie-segmentation:256x144 --overlay_mode`
+- `Archive run:` `python3 main.py --model /path/to/custom-model.tar.xz`
 - `Success looks like:` Visualizer exposes `Video` and `Detections`, and the pipeline runs until `q` is pressed
-- `Common failure meaning:` model slug unavailable for platform, private model auth missing, or selected model violates the single-input/single-output assumptions
+- `Common failure meaning:` model identifier unavailable for platform, private model auth missing, or selected model violates the single-input/single-output assumptions
diff --git a/neural-networks/generic-example/README.md b/neural-networks/generic-example/README.md
index 2e3e761a8..1d64a3fd4 100644
--- a/neural-networks/generic-example/README.md
+++ b/neural-networks/generic-example/README.md
@@ -1,7 +1,7 @@
 # Generic Example
 
 We provide here an example for running inference with a **single model** on a **single-image input** with a **single-head output**.
-The example is generic and can be used for various single-image input models from the [Model ZOO](https://models.luxonis.com).
+The example is generic and can be used for various single-image input models from the [Model ZOO](https://models.luxonis.com) or HubAI, a model descriptor/YAML, or a local `.tar.xz` `NNArchive`.
 
 ## Usage
 
@@ -13,7 +13,7 @@ Here is a list of all available parameters:
 
 ```
 -m MODEL, --model MODEL
-                      HubAI model reference. (default: luxonis/yolov6-nano:r2-coco-512x288)
+                      Model Zoo/HubAI model identifier, model YAML/descriptor, or local .tar.xz NN archive. (default: luxonis/yolov6-nano:r2-coco-512x288)
 -d DEVICE, --device DEVICE
                       Optional name, DeviceID or IP of the camera to connect to. (default: None)
 -fps FPS_LIMIT, --fps_limit FPS_LIMIT
@@ -21,7 +21,7 @@ Here is a list of all available parameters:
 -media MEDIA_PATH, --media_path MEDIA_PATH
                       Path to the media file you aim to run the model on. If not set, the model will run on the camera input. (default: None)
 -api API_KEY, --api_key API_KEY
-                      HubAI API key to access private model. Can also use 'DEPTHAI_HUB_API_KEY' environment variable instead. (default: )
+                      HubAI API key for private HubAI access. Can also use 'DEPTHAI_HUB_API_KEY' environment variable instead. (default: )
 -overlay OVERLAY_MODE, --overlay_mode
                       If passed, overlays model output on the input image when the output is an array (e.g., depth maps, segmentation maps). Otherwise, displays outputs separately.
 ```
@@ -67,6 +67,13 @@ python3 main.py \
 
 And this will run an instance segmentation model.
 
+```bash
+python3 main.py \
+    --model /path/to/custom-model.tar.xz
+```
+
+This will run a local `NNArchive` directly instead of fetching a model from the Model Zoo.
+
 ## Standalone Mode (RVC4 only)
 
 Running the example in the standalone mode, app runs entirely on the device.
diff --git a/neural-networks/generic-example/main.py b/neural-networks/generic-example/main.py
index 36ea4e3df..733bc0ecb 100644
--- a/neural-networks/generic-example/main.py
+++ b/neural-networks/generic-example/main.py
@@ -23,10 +23,17 @@
     print("Creating pipeline...")
 
     # model
-    model_description = dai.NNModelDescription(f"yolov6_nano_r2_coco.{platform}.yaml")
-    if model_description.model != args.model:
-        model_description = dai.NNModelDescription(args.model, platform=platform)
-    nn_archive = dai.NNArchive(dai.getModelFromZoo(model_description))
+    default_description = dai.NNModelDescription(f"yolov6_nano_r2_coco.{platform}.yaml")
+
+    if args.model.endswith(".tar.xz"):
+        nn_archive = dai.NNArchive(args.model)
+    else:
+        model_description = (
+            default_description
+            if default_description.model == args.model
+            else dai.NNModelDescription(args.model, platform=platform)
+        )
+        nn_archive = dai.NNArchive(dai.getModelFromZoo(model_description))
 
     # media/camera input
     input_node = create_input_node(
diff --git a/neural-networks/generic-example/utils/arguments.py b/neural-networks/generic-example/utils/arguments.py
index 5a5764e89..cd412a9a9 100644
--- a/neural-networks/generic-example/utils/arguments.py
+++ b/neural-networks/generic-example/utils/arguments.py
@@ -7,16 +7,16 @@ def initialize_argparser():
         formatter_class=argparse.ArgumentDefaultsHelpFormatter
     )
     parser.description = (
-        "General example script to run any model available in HubAI on DepthAI device. \
-        All you need is a model slug of the model and the script will download the model from HubAI and create \
-        the whole pipeline with visualizations. You also need a DepthAI device connected to your computer. \
+        "General example script to run a single-model DepthAI pipeline from a Model Zoo/HubAI model identifier, \
+        a model descriptor/YAML, or a local .tar.xz NN archive. The script creates the pipeline and visualizations \
+        for a connected OAK device. \
         If using OAK-D Lite, please set the FPS limit to 28."
     )
 
     parser.add_argument(
         "-m",
         "--model",
-        help="HubAI model reference.",
+        help="Model Zoo/HubAI model identifier, model YAML/descriptor, or local .tar.xz NN archive.",
         default="luxonis/yolov6-nano:r2-coco-512x288",
         type=str,
     )
@@ -51,7 +51,7 @@ def initialize_argparser():
     parser.add_argument(
         "-api",
         "--api_key",
-        help="HubAI API key to access private model. Can also use 'DEPTHAI_HUB_API_KEY' environment variable instead.",
+        help="HubAI API key for private Model Zoo access. Can also use 'DEPTHAI_HUB_API_KEY' environment variable instead.",
         required=False,
         default="",
         type=str,