Skip to main content

Cyberwave is in Private Beta.

Request early access to get access to the Cyberwave dashboard.

cw.models is the unified entry point for both the model catalog and runtime inference.
This page covers the SDK surface only. For model concepts, the playground, and structured actions, see the Models feature pages linked at the bottom.

Browse the catalog

# All models visible in your workspace
for m in cw.models.list():
    print(f"{m.slug:55s}  {m.deployment}  sdk_load_id={m.sdk_load_id}")

# Filter by deployment target (server-side)
edge_models  = cw.models.list(deployment="edge")
cloud_models = cw.models.list(deployment="cloud")

# Client-side shorthands (ANDed): e.g. edge + image-capable rows
edge_vision = cw.models.list(filters=["edge", "image"])

public_models = cw.models.list_public()

# Get a single record
m = cw.models.get("acme/models/yolov8-nano")
m = cw.models.get_by_uuid("3f2a1b4c-…")

Load and predict

from PIL import Image

# Pass a catalog entry directly — the SDK resolves the right load key
entry = cw.models.list(deployment="edge")[0]
model = cw.models.load(entry)

# Or pass a string (filename, slug, or UUID)
model = cw.models.load("yolo26n.pt")
pred  = model.predict(Image.open("frame.jpg").convert("RGB"), confidence=0.25)

print(pred.describe())
for d in pred:
    print(d.label, d.confidence, d.bbox)

Delete a catalog record

cw.models.delete("model-uuid-here")

Predict on a live twin frame

Combine with Frames & Cameras to run a model on the latest frame from a twin:
arr = robot.get_frame("numpy")          # latest frame as a numpy array
pred = model.predict(arr, confidence=0.25)

ML Models

What ML Models are and how cloud/edge inference works.

Model Playground

Test models interactively and grab integration snippets.

Structured Actions

Preset catalog of structured tasks.

Speech to Text

Whisper-based transcription.