making cpsam_v2 default

Author: carsen-stringer

README.md

@@ -32,7 +32,7 @@ Example notebooks:
 
 New Cellpose4 models added! We have included new Cellpose models based on [DINOv3](https://github.com/facebookresearch/dinov3), `cpdino` and `cpdino-vitb`. 
 We have also updated the CellposeSAM model (`cpsam_v2`), and it now predicts fewer spurious masks in low-contrast regions. 
-To use the new models, `pip install cellpose[dino] --upgrade`, and find more details [here](https://cellpose.readthedocs.io/en/latest/models.html).
+To use the new models, `pip install cellpose --upgrade` and `pip install git+https://github.com/facebookresearch/dinov3`, and find more details [here](https://cellpose.readthedocs.io/en/latest/models.html).
 
 :triangular_flag_on_post: All Cellpose models are trained on data that is licensed under **CC-BY-NC**. The Cellpose annotated dataset is also CC-BY-NC.
 
@@ -105,7 +105,8 @@ If you are using a GPU, make sure its drivers and the cuda libraries are correct
 3. Create a new environment with `conda create --name cellpose python=3.12`. We recommend python 3.12, but python 3.9 and 3.11 will also work.
 4. To activate this new environment, run `conda activate cellpose`
 5. (option 1) To install cellpose with the GUI, run `python -m pip install cellpose[gui]`.  If you're on a zsh server, you may need to use ' ': `python -m pip install 'cellpose[gui]'`.
-6. (option 2) To install cellpose without the GUI, run `python -m pip install cellpose[dino]`. 
+6. (option 2) To install cellpose without the GUI, run `python -m pip install cellpose`. 
+7. Install the DINOv3 package to use the CellposeDINO models with `python -m pip install git+https://github.com/facebookresearch/dinov3`.
 
 To upgrade cellpose (package [here](https://pypi.org/project/cellpose/)), run the following in the environment:
 

cellpose/cli.py

@@ -65,9 +65,9 @@ def get_arg_parser():
 
     # model settings
     model_args = parser.add_argument_group("Model Arguments")
-    model_args.add_argument("--pretrained_model", required=False, default="cpdino",
+    model_args.add_argument("--pretrained_model", required=False, default="cpsam_v2",
                             type=str,
-                            help="path to model for segmentation or starting training, or builtin model name: cpdino, cpsam_v2, cpdino-vitb, or cpsam")
+                            help="path to model for segmentation or starting training, or builtin model name: cpsam_v2, cpdino, cpdino-vitb, or cpsam")
     model_args.add_argument(
         "--add_model", required=False, default=None, type=str,
         help="model path to copy model to hidden .cellpose folder for using in GUI/CLI")

cellpose/models.py

@@ -28,7 +28,7 @@
 _MODEL_DIR_DEFAULT = Path.home().joinpath(".cellpose", "models")
 MODEL_DIR = Path(_MODEL_DIR_ENV) if _MODEL_DIR_ENV else _MODEL_DIR_DEFAULT
 
-MODEL_NAMES = ["cpdino", "cpsam_v2", "cpdino-vitb", "cpsam"]
+MODEL_NAMES = ["cpsam_v2", "cpdino", "cpdino-vitb", "cpsam"]
 
 MODEL_LIST_PATH = os.fspath(MODEL_DIR.joinpath("gui_models.txt"))
 
@@ -89,7 +89,7 @@ class CellposeModel():
         nclasses (int): Number of classes in the model.
         nbase (list): List of base values for the model.
         net (CPnet): Cellpose network.
-        pretrained_model (str): Name of pretrained model ("cpdino", "cpsam_v2", etc) or path to pretrained cellpose model.
+        pretrained_model (str): Name of pretrained model ("cpdino", "cpsam_v2", etc) or path to pretrained cellpose model. Default is "cpsam_v2".
         backbone (str): Type of network ("default" is the standard res-unet, "transformer" for the segformer).
 
     Methods:
@@ -101,7 +101,7 @@ class CellposeModel():
 
     """
 
-    def __init__(self, gpu=False, pretrained_model="cpdino", model_type=None,
+    def __init__(self, gpu=False, pretrained_model="cpsam_v2", model_type=None,
                  diam_mean=None, device=None, nchan=None, use_bfloat16=True):
         """
         Initialize the CellposeModel.
@@ -149,7 +149,7 @@ def __init__(self, gpu=False, pretrained_model="cpdino", model_type=None,
                 models_logger.warning(
                     f"pretrained model {pretrained_model} not found, using default model"
                 )
-                pretrained_model = cache_model_path("cpdino")
+                pretrained_model = cache_model_path("cpsam_v2")
 
         self.pretrained_model = pretrained_model
         dtype = torch.bfloat16 if use_bfloat16 else torch.float32 

docs/models.rst

@@ -8,8 +8,8 @@ Built-in models
 
 We have four built-in models available:
 
-* ``cpdino``: this is the CellposeDINO model released in June 2026 using the DINOv3-ViTL backbone
 * ``cpsam_v2``: this is the CellposeSAM model released in June 2026 using the SAM-ViTL backbone, it includes a fix in the training for low contrast regions
+* ``cpdino``: this is the CellposeDINO model released in June 2026 using the DINOv3-ViTL backbone
 * ``cpdino-vitb``: this is the CellposeDINO model released in June 2026 using the DINOv3-ViTB backbone (smaller model)
 * ``cpsam``: this is the original CellposeSAM model released in April 2025 using the SAM-ViTL backbone