⚙️
📱
⚡
🧠
🔋
📡
Now in Public Beta

2Device

Device-aware compute orchestration.
Run multiple ML models on mobile — without crashing.

Start Building See How It Works
0
Memory Cascade Levels
0
Profiling Latency
0
Crash Prevention
main.cpp
// Initialize 2Device runtime 2d_config_t config = 2d_default_config(); config.max_memory_mb = 512; config.thermal_limit = 0.85; 2d_runtime_t* rt = 2d_create(&config); // Load models — 2Device handles the rest 2d_load_model(rt, "whisper-tiny", 2D_PRIORITY_HIGH); 2d_load_model(rt, "llama-7b-q4", 2D_PRIORITY_MED); 2d_load_model(rt, "moondream-v2", 2D_PRIORITY_LOW); // Intelligent orchestration — zero crashes 2d_run(rt);
Scroll to explore
Core Capabilities

Everything your device
needs to think

Foundational systems that transform any mobile device into an intelligent compute platform.

🧠

Device Profiler

Real-time hardware fingerprinting. Detects compute capabilities, memory topology, and accelerator availability — all in under 12ms.

💾

Adaptive Memory Engine

Multi-tier resource management with deep OS-level integration on iOS and Android. Preemptive eviction before the system intervenes. Zero OOM kills.

🌡️

Thermal Intelligence

Continuously adapts workload intensity based on real-time thermal state. Graceful scaling before the OS throttles you.

📊

Dynamic Resource Scheduler

Intelligently routes workloads across threads and GPU cores in real-time. Run 3+ models concurrently without contention.

🔊

Audio Pipeline

Full-stack voice processing with echo cancellation, activity detection, gain control, and noise suppression — all on-device, real-time, hardware-accelerated.

🛡️

Secure OTA Policies

Cryptographically signed behavior policies delivered via our global edge network. Update limits and rules without app store review.

Integration

Three lines of code.
Infinite intelligence.

From install to production in minutes — not weeks.

1

Link the SDK

Static library + two headers. CMake, CocoaPods, or drop the XCFramework into Xcode. No runtime dependencies.

2

Declare your models

Tell 2Device what you want to run and at what priority. It profiles the device and builds an optimal execution plan.

3

Call 2d_run()

Intelligent resource management, thermal awareness, and workload orchestration activate automatically. Ship with confidence.

4

Monitor & adapt

Optional telemetry and over-the-air policy updates. Tune model priorities and resource limits without shipping a new build.

Benchmarks

Performance that
speaks for itself

0
Crash Prevention
0
Profiling Speed
0
Concurrent Models
0
SDK Binary Size
Under the Hood

Built for the
real world

2Device
Runtime
🔍
💾
🌡️
📊
🔊
🛡️

⚡ Universal API Surface

Single C-compatible interface — link from Swift, Kotlin, C++, Rust, or WASM. Drop into any project in minutes, not weeks.

📱 Deep OS Integration

Direct hardware queries on every major mobile platform. No wrappers, no bridging overhead — true system-level awareness.

🚀 Hardware Acceleration

Native SIMD and platform-specific acceleration paths. Every computation runs at the maximum speed your hardware allows.

🌐 Policy/Engine Split

Engine ships as a sealed binary. Behavior rules update over the air via our tamper-proof edge network — no app store review needed.

Plans

Simple pricing.
Scale when ready.

Open Source

Free

$0

Full SDK. Attribution required.

  • All runtime features
  • Community support
  • Telemetry opt-in
  • Free commercial license
Get Started
Most Popular

Indie

$49/mo

No attribution. Priority support.

  • Everything in Free
  • Remove attribution
  • Email support
  • Analytics dashboard
Start Free Trial
Teams

Pro

$299/mo

Dashboard, policy CDN, SLA.

  • Everything in Indie
  • Edge policy delivery
  • Custom dashboards
  • Priority SLA
Contact Sales

Ready to build
without limits?

Start running ML models on mobile today. Zero crashes, zero compromises.

Get the SDK