Cactus

About Cactus

A hybrid edge-cloud inference engine with OpenAI-compatible APIs, custom ARM kernels, and 1-4-bit quantization, so models can run locally on mobile hardware. Native bindings cover Swift, Kotlin, Flutter, React Native, Python, and Rust. Fills the on-device slot in an otherwise cloud-only inference field.

Run text, speech, and vision models on-device on phones and wearables.

Similar tools

OpenRouterOne API for hundreds of models, with fallback routing.GroqRun open models at extreme speed on custom hardware.Together AIHost and fine-tune open models via a fast API.ReplicateRun and deploy any model with one API call.Hugging FaceThe hub for open models, datasets, and demos.Fireworks AIFast, production inference for open models.