Your edge. Supercharged.
Connect with Solana. Create with a real NVIDIA GPU.
Intelligence.
In your hands.
Sign in with Solana. Run on NVIDIA Jetson Orin.
Real CUDA inference, from your own edge.
NVIDIA Jetson Orin Nano
CLAWD runs inference on an NVIDIA GPU.
Independent project; no NVIDIA sponsorship or endorsement.
NVIDIA hardware, live
Waiting for device sampleCUDA activity
Measured on the Orin · sampled every 5 seconds
Inference performance
Q4_K_MLast completed generation · llama.cpp
- Request latency
- —
- GPU slot
- Checking
- Requests / failures¹
- —
- FP32 capacity estimate²
- —
¹ Since request tracking was enabled. ² Calculated FP32 ceiling at the current GPU clock (512 cores × 2 × GHz); not measured TFLOPS or Tensor Core TOPS. Board power is unavailable without sensor read permission.
Put your GPU to work
OpenAI-compatible · Streaming responsesPlayground CLAWD
Qwen 2.5 · 0.5BWhat will you build?
Ask a question and watch your GPU respond.
Your Solana access
Your wallet. Your compute.
Connect any Solana wallet with message signing. No token balance required.
Connect a wallet to unlock inference.
Sign with Solana
Prove wallet ownership with a message. Receive your own revocable API key.
Run on your GPU
Qwen inference on 512 CUDA cores. Stream every response from real hardware.
See every pulse
Convex streams GPU telemetry to your dashboard, straight from the device.
Your next API call runs here.