kitio.aiAdded Oct 28, 2026
Carton
FreeOpen-source library to run ML models from many programming languages
What is Carton?
Carton is an open-source runtime that lets applications load and run machine learning models through one API, regardless of the framework the model was built in. It targets developers who want inference code that does not depend on PyTorch or TensorFlow directly.
A model is packed by wrapping the original file with metadata about its framework and version into a .carton zip, without converting it. When loaded, Carton reads that metadata, picks the right runner, fetches it if needed, and passes your inference calls through to the framework.
The core is written in Rust with bindings for Python, Rust and C, while JavaScript, TypeScript, Java, Go, Swift and several other languages are marked as work in progress. A community model registry is available alongside the docs and quickstart.
Features
- Framework-agnostic inference API
- Model packing without conversion
- Automatic runner selection
- Rust core with language bindings
- Community model registry
Tags
Releases
+ Submit new release- Listed
Carton was added to the directory.
Organization
Carton
Listed by kitio.ai · unverified owner
Pricing
- Pricing model
- Free
- Paid options from
- 100% free
Sponsored
Your AI tool here
A Featured launch puts your tool in this spot and on the home page for a week.
Top Carton alternatives
View allModels & APIs
The main hub for open models, datasets and AI demo apps
Models & APIs
Run and deploy open-source AI models through a simple API
Models & APIs
Fast inference and fine-tuning platform for open models
Models & APIs
Cloud for running, fine-tuning and training open-source models
If you liked Carton
Low-cost serverless API for hundreds of open-source models
PaidModels & APIsMade in United StatesCompute, RL training, evals and inference for building your own models
PaidModels & APIsMade in United StatesOpen-source model serving and a managed platform for AI inference
FreemiumModels & APIsMade in United States
Pros and cons
Pros
- Free and open source
- No error-prone model conversion
- Same API across frameworks
Cons
- Many language bindings still in progress
- Small community compared to ONNX tooling
- Developer-only, no hosted service
Prompts & results
No prompts yet. Add your own prompt and output to help others use this tool.
Share a prompt
Reviews
Q&A
What is Carton?
Carton is an open-source runtime that lets applications load and run machine learning models through one API, regardless of the framework the model was built in. It targets developers who want inference code that does not depend on PyTorch or TensorFlow directly.
Does Carton change my original model file?
No. It stores the original model unchanged inside a zip with a small metadata file describing the framework and version it needs.
Which languages can I use Carton from?
Python, Rust and C are available, and bindings for languages such as JavaScript, TypeScript, Java, Go, Swift and C# are listed as work in progress.
How much does Carton cost?
Carton is 100% free.
Is there a free version of Carton?
Yes. Carton is completely free to use.
What are the main features of Carton?
Framework-agnostic inference API, Model packing without conversion, Automatic runner selection, Rust core with language bindings, Community model registry.
Who makes Carton?
Carton is made by Carton.
What are the best alternatives to Carton?
Popular alternatives include Hugging Face, Replicate, Fireworks AI.