I can only answer that for the inter-species range for Corynebacterium. There the difference can be huge. So far, the lowest was for C. kutscheri with 46.46% (see link) and the highest was C. sphenisci with  74.73% (not published yet). The highest published value when you exclude T. otitidis (71.4%, is a Corynebacterium in my book) is for C. nuruki (69.5%). That gives you almost 30% difference.

Within a species, this cannot go above 4-5% difference, given that the ANI cutoff for a species is assumed to be 95-96%. And this would assume that every A/T is exchanged for a G/C. That is rather unlikely (although nature has a habit to do unlikely things). So I would not expect more than 0.5-1% in general.

http://genomea.asm.org/content/3/3/e00571-15.short

http://www.ncbi.nlm.nih.gov/pmc/articles/PMC3147667/

At present, there is no good, universally accepted answer to your question. For this reason, many authors avoid the use of the "species" term when talking about bacteria, using the OTU term instead and an arbitrary cutoff value. As data accumulate and knowledge increases, cutoff values will become less arbitrary - and microbiologists will be able to decide on a definition of species for the prokaryotes. Meanwhile, I advise you to perform a thorough literature search, to see what cutoff values other authors in your area/bacterial group are using.

Hope this helps.

for Escherichia genomes, the G + C genomic content ranges between 50.4% and 50.8% (citation in Mann et Chen 2010)

http://www.sciencedirect.com/science/article/pii/S0888754309002080

In general, a difference of less than 5% in GC content is adequate to assign two microorganisms to the same species and the difference by more than 10% GC should not be classified within the same genus (Goodfellow and O'Donnell, 1993)