Developer Tool
Browser Diagnostic
Check your browser's WebGPU support and get a realistic estimate of local inference performance before integrating the MeshInfer SDK.
WebGPU Not Available
Your browser doesn't support WebGPU. Try Chrome 121+, Edge 121+, or Safari 17.4+. MeshInfer will fall back to WASM SIMD inference (~6–12 tok/s) or route to the mesh/cloud.
Hardware Profile
CPU Cores4 logical cores
Device Memory2 GB
PlatformLinux
User AgentNot;A=Brand 24, Chromium 128
Connection4g · 10 Mbps
WASM SIMDSupported
Estimated Performance
Based on your detected GPU tier: No WebGPU — WASM fallback
Llama-3.2-1B (Q4)VRAM ~0.8GB
WASM fallback10 tok/s
Phi-3.5-mini (Q4)VRAM ~2.2GB
WASM fallback6 tok/s
Qwen-2.5-7B (Q4)VRAM ~4.5GB
WASM fallback3 tok/s
What this means
- WASM path handles 1B models acceptably. Larger models will route to mesh/cloud.
- Upgrade your browser to unlock WebGPU for 3–6× faster local inference.
Compute Benchmark
Runs a 256×256 matrix multiply to measure raw throughput. Takes ~2 seconds.
Predicted Route Mix
How MeshInfer would likely route traffic for this device, given detected capabilities.
Llama-3.2-1B (Q4)
60%
20%
20%
Phi-3.5-mini (Q4)
15%
45%
40%
Qwen-2.5-7B (Q4)
30%
70%
LocalMeshCloud
Ready to integrate?
These results reflect what the MeshInfer SDK's capability probe detects at runtime. The router uses this data — plus network conditions and model availability — to pick the optimal route for every request.