WebGPU not available.
This tool requires WebGPU. Please use Chrome 113+ or Edge 113+.
If you're on a supported browser, try enabling
chrome://flags/#enable-unsafe-webgpu
.
Model Setup
Model
(downloaded & cached from HuggingFace on first use - loaded in browser memory)
Load Model
Switch model
Base URL
(must serve /v1/completions with logprobs, e.g. llama.cpp, vLLM, LM Studio)
API Key
(optional)
Model name
Tokens to show:
10
Temperature:
1.0
Raw text
Chat template
Starting point
System prompt
You are a helpful coding assistant.
User / code
Get Next Token
Explorer