Last updated 07-30-2026
Category:
Reviews:
Join thousands of AI enthusiasts in the World of AI!
Sinkove
Sinkove turns clinical prompts into synthetic radiology datasets for research teams. Diffusion models generate configurable medical images by disease subtype, imaging protocol, and scan description, then package them for download through a web dashboard, Python SDK, or REST API. The workflow targets regulated medical image generation rather than open-ended creative output.
Most general image generators output creative visuals without clinical guardrails. Sinkove narrows the scope to regulated medical imaging workflows: you pick from public or organization-specific models, preview a single synthetic scan before committing, and generate full datasets capped around 1000 images per run. Private model training requires contacting the Sinkove team, which signals the product is aimed at research teams rather than casual creators.
The platform fits biomedical researchers, clinical trial coordinators, and healthcare data scientists who need balanced imaging cohorts without waiting on multi-year collection cycles. Organization accounts support team member invites, API keys, and dataset state tracking from PENDING through READY.
Full dataset runs scale up to about 1000 synthetic medical images per generation
Python SDK installs with pip install sinkove-sdk and requires Python 3.12+
Dataset states track PENDING, STARTED, READY, and FAILED through the API
Public open-source models are available to every Sinkove account
Preview generation creates one sample image before you commit to a full dataset
REST API requests authenticate with Bearer tokens from Profile > API Keys
Public open-source models are available without a sales conversation.
Preview a single synthetic scan before generating a full dataset.
Python SDK and REST API support automated research pipelines.
Organization workspaces include team invites and shared model access.
Private model access requires emailing [email protected].
The Python SDK requires Python 3.12 or newer.
Full dataset generation can take minutes to hours depending on batch size.
What does Sinkove generate?
Sinkove generates synthetic medical imaging datasets using diffusion models. You configure prompts such as chest X-ray findings, choose a model ID, and download ZIP archives through the web app or Python SDK.
How many images can Sinkove create per dataset?
Sinkove supports full dataset generation up to about 1000 synthetic images per run. The quick start guide also lets you generate a single preview sample before starting a full batch.
Does Sinkove offer a Python SDK?
Yes. Sinkove publishes the sinkove-sdk package on PyPI, installable with pip install sinkove-sdk. The SDK requires Python 3.12+, an organization UUID, and a SINKOVE_API_KEY environment variable.
How do you access private models on Sinkove?
Private models are limited to your organization on Sinkove. The models documentation directs teams that need private or custom-trained models to email [email protected] for access.
What dataset states does Sinkove use?
Sinkove datasets move through PENDING, STARTED, READY, and FAILED states. The Python SDK exposes dataset.ready and dataset.state so scripts can wait for completion before downloading output ZIP files.
How do you authenticate Sinkove API requests?
Sinkove REST API calls use Authorization: Bearer headers with keys generated under Profile > API Keys in the dashboard. The Python SDK reads the same key from the SINKOVE_API_KEY environment variable.