Active Simulations
No active simulations
All Simulations — Today
Loading...
Live Activity Feed
No activity yet
Run Lanes — Night One
🔴 Large Lane
High-capacity models — slower, deeper output
qwen3.5-35b-a3b
google/gemma-4-26b-a4b
🔵 Small Lane
Smaller models — faster, higher throughput
qwen/qwen3.5-9b
unsloth/gemma-4-e4b-it
lmstudio-community/gemma-4-e4b-it
Same prompt runs in both lanes simultaneously.
When small lane finishes → pull next from queue.
Large lane runs at its own pace.
When small lane finishes → pull next from queue.
Large lane runs at its own pace.
Prompt Queue — 15 Scenarios
Loading queue...
Hardware Thermals — dodo (M3 Ultra)
CPU Temp
—
GPU Temp
—
CPU Power
—
LLM Temp
0.7
Hardware thermals require the overnight benchmark daemon to be running on dodo. LLM temperature is the parameter sent to the model.