Our new, fastest-ever Cactus v1 SDK is here!

Learn more →
Y CombinatorOxford Seed FundGoogle for Startups

The fastest way to deploy mobile AI

Deploy AI models locally on smartphones - in Flutter, React Native, and Kotlin Multiplatform.

Minimize latency, guarantee privacy, and decrease server costs.

Get StartedDashboard3.0k+ stars

<50ms

Time to First Token

Up to 300

Tokens / second

Zero

Data Leaves the Device

Designed for the Edge

Unified cross-platform SDK

Offline-ready
Perfect for unreliable networks or internet-disabled devices.
Private
On-device inference by default. No data transmission and complete user privacy.
Multimodal
Deploy language, vision, and speech models through a unified framework.
Cloud fallback
Fall back to cloud inference if needed for longer or asynchronous tasks.
Agentic
Augment your workflows with built-in mobile tool calling.
Native Support
iOS xcframework and Android JNILibs for seamless native integration

Platform Support

Get started with your preferred framework

Flutter
Dart package for Flutter apps
Quick Start →
React Native
NPM package for React Native
Quick Start →
Kotlin Multiplatform
Kotlin package for KMP
Quick Start →

Built-in Telemetry

One-line initialization with a CACTUS_TELEMETRY_TOKEN.

Track device engagement in real time

Monitor user activity, model usage, device performance, and inference types. Understand your user patterns without additional setup or configuration.

Get instant visibility into device-level metrics, inference throughput, latency, and user engagement.

Daily active devices
Daily active devices across your Cactus projects
Error rate
Daily error rate across your events

Optimize workflow performance

Capture error rates across your deployments. Identify problematic patterns or workflow performance degradation in real-time.

Run out-of-the-box analytics to ensure your AI features remain reliable and performant.

Start tracking →

Agent Builder Canvas

Create complex workflows on a simple interface

Deploy →

Performance benchmarks

Real-world performance data on popular consumer devices

Tokens per Second
Real-world decode performance measured through our demo apps
394MB

Frequently Asked Questions

Everything you need to know about deploying AI on mobile

Try demo apps

Experience Cactus SDK in action with our demo applications

Download on App StoreGet it on Google Play

Ready to find your Edge?

Join thousands of developers building the future of mobile AI

Contact usRead DocumentationJoin Discord