AETHER

Inference Benchmarking

Transformers.js v3 + WebGPU

One-Click Automated Benchmarking Suite

Execute all 9 testing matrices sequentially (Short, Medium, and Long text vs. Client WASM, Client WebGPU, and Server). This generates a complete grouped crossover visual to isolate the tipping point.

1. Select Benchmark Payload
Character Count: 385Approx. Words: 56
2. Manual Target Override
Targeted Run Diagnostics
RTT Latency—Warm Median
Pure Inference—Warm Compute
Transit Overhead—Network/Worker Lag
Cold Init—Graph Compiler
Execution Logs Console
Ready to test. Click "Run Automated Suite" above to benchmark everything.

Grouped Payload Crossover Analysis (Warm RTT)

Live Railway Cloud Network Latency
No batch measurements recorded. Run the Automated Suite above to compile crossover metrics.