Metagenomic binning enables the in-depth characterization of microorganisms. To improve the resolution and efficiency of metagenomic binning, BASALT (Binning Across a Series of AssembLies Toolkit), a novel binning toolkit was present in this study, which recovers, compares and optimizes metagenomic assembled genomes (MAGs) across a series of assemblies from short-read, long-read or hybrid strategies. BASALT incorporates self-designed algorithms which automates the separation of redundant bins, elongate and refine best bins and improve contiguity. Evaluation using Critical Assessment of Metagenome Interpretation (CAMI) datasets at low (132 genomes) and medium (596 genomes) complexities revealed that BASALT auto-binning gained up to 10 times of shared MAGs with better quality when pairwise comparing to DASTool, VAMB and metaWRAP, resulting in up to 51% more MAGs obtained than other binners. Using BASALT, a case-study analysis of a Salt Lake sediment microbial community from northwest arid region of China was performed, resulting in 426 non-redundant MAGs, including 352 bacterial and 69 archaeal MAGs which could not be assigned to any known species from GTDB (ANI < 95%). In addition, two Lokiarchaeotal MAGs that belong to superphylum Asgardarchaeota were observed from Salt Lake sediment samples. This is the first time that candidate species from phylum Lokiarchaeota was found in the arid and deep-inland environment, filling the current knowledge gap of earth microbiome. Overall, BASALT is proven to be a robust toolkit for metagenomic binning, and more importantly, expand the Tree of Life.