Strata 2 beta running at half speed?

Explain Like I'm Normal

Users are reporting significant performance regressions in the Strata 2.0 beta, with inference speeds dropping from 155 t/s to under 100 t/s on high-end dual GPU setups. This highlights the volatility of local quantization frameworks where new 'optimizations' can often break specific hardware configurations. Builders should wait for the stable release before migrating production workflows.