Local AI Health
All inference runs on your device. Zero cloud. Zero API cost.
AI Pipeline Status
Models loaded into local memory — no network calls, no APIs
Checking model states...
Inference Performance
Real-time latency from local model execution
Connecting...
AI Power Management
Suspend pauses model inference to save battery. Resume restarts instantly.
When suspended, all model inference stops and your device's GPU and CPU are freed. Resuming picks up exactly where you left off — no model reload needed.
Decentralized AI Network
Share your device's AI power with the Cortex network
Checking network...
P2P inference delegation via Holepunch. Decentralized AI without servers. Your models stay on your device — you just lend compute to peers who need it.
On-Chain Attestation
Pipeline metrics recorded on Solana devnet after every voice command
What gets attested
- Intent hash of each voice command
- Embedding similarity scores for semantic verification
- Inference timing for each pipeline stage
- Model identifiers (Whisper, LLM, TTS versions)
Every inference run produces a cryptographic receipt on-chain. Zero data leaves your device — only the hashes and timing are recorded, proving the pipeline ran without revealing your input.
Available Models
QVAC model registry — every model your device can run locally
Fetching registry...