L2S1 ↗

REAL INFERENCE / IN YOUR BROWSER

This browser
makes the decision.

Run Qwen3 0.6B ONNX with WebGPU.
Load the model, then inspect decisions and candidate scores for your own text.

01 · Prepare the model

onnx-community/Qwen3-0.6B-ONNX ↗ · Apache 2.0

Checking support

Checking WebGPU support…

02 · Input and decision policy

Edit questions and candidates

binary / choice / ordinal · Up to 8 questions and 26 candidates per question. Your own IDs and candidate criteria are used.

Top candidate probability ≥ 1 − input rate.

Custom abstention and inference failure messages

Edit the message displayed for each failure code.

Checking WebGPU support…

03 · Real model results

Load the model, then analyze