Adoptive cell therapy including chimeric antigen receptor (CAR) T cells targeting CD19 has been approved by FDA to treat B cell-derived malignancies with remarkable success. The success has not yet been expanded to treating Acute Myeloid Leukemia (AML). We previously showed that a nanobody and single-chain fragment variable (scFv) CD13 (Nanobody)/TIM-3 (scFv) directed bispecific split CAR (bissCAR) T cells, while effective in eliminating AML in preclinical models, also caused substantial toxicity to human hematopoietic stem cells (HSCs) and other lineages. To maintain the bissCART specificity and efficacy, yet reduce toxicity to normal cells including HSCs, we generated new anti-TIM-3 nanobodies and constructed new cognate nanobodies-directed CD13/41BB and TIM3/CD3zeta nbiCARTs. The resultant nbiCARTs showed strong antitumor activity to CD13/TIM3 positive leukemic cells in vitro and in preclinical models. Importantly, the 3rd generation of nbiCARTs had little toxicity to human bone marrow-derived colony forming progenitors ex vivo and the human HSCs in mice with a humanized immune system. Together, the current studies generated novel and 3rd G CD13/TIM-3 nbiCARTs that displayed stronger antitumor activity yet minimal toxicity to normal tissues like HSCs that express a moderate level of CD13, paving the way to further evaluate the novel CD13/TIM-3CARTs in treating aggressive and refractory AML in clinical studies.