Problem
For drug testing, "animal model" medical research has no predictive value; an animal model is just as likely to give the wrong answer about what will happen in human subjects as it is likely to give useful reliable guidance to a Phase 1 trial. See, for example, this analysis, or this one. And if you don't believe the information in these documents, in 2019 the Environmental Protection Agency announced that it will soon end the practice of animal testing for determining the health effects of chemicals, because the testing is "unreliable and misleading."
Yet there is virtually always some level of suffering inflicted on the animals tested.
Question: If animal models are not predictive for drug testing, how is a study that poisons several hundred dogs or cats (as a measure of toxicity) morally different from Sam mistreating Skippy?