AI SpeedtestTest your AI performance
Measure how fast your device runs AI models in your browser.
The first run downloads about 400 MB of model files, which are cached for future tests.
Graphics processor
Detecting your GPU…
- Text generation
- —tok/s
- Prompt processing
- —tok/s
- Embeddings
- —tok/s
- Time to first token
- —ms
Device and test details
- Graphics processor
- Detecting…
- Logical processors
- Not exposed
- Reported memory
- Not exposed
- Storage buffer limit
- Not exposed
- Float16 support
- Checking…
- Language model
- SmolLM2 · 360M
- Embedding model
- Arctic Embed · 33M
- Runtime
- WebLLM 0.2.85