Some say F5, some say 6, now 7... What matters only is the selection the breeder makes. It indicates the generation it does not dictate stability. Theoretically an F3 can be more stable and more uniform than an F10 generation.
All F3 offspring can already be entirely homozygous for all the visible traits and as stable as it gets for both pheno and genotype. F2 already contains plants with homozygous genotypes, if you pick and mate two of those you get a homozygous pure line true bred super stable strain. The traits inherit independently (segregated inheritance), so the number of plants that have all the traits you want and in homozygous form are very rare and it would obviously require a lot of luck.
For example, the classic example:
(S is dominant smooth round trait, s is recessive wrinkled trait, Y is yellow dominant trait, y is recessive green color trait. Side note: dominant and recessive is relatively, eg. red could be dominant over yellow, making yellow recessive... yes that simple).
If you select a plant with the genetic make up of the top one in the F2 gen, the SSYY, and cross that with another plant with that same genetic make up, all the F3 (and F4 and on) will be SSYYxSSYY = SSYY as in all smooth round yellow.
Same thing for the one below it, pick two F2 plants with SSyy and all the offspring will be SSyy.
Same for ssyy x ssyy and ssYY x ssYY.
If the goal is smooth round yellow as the only phenotype even crossing SSYY with ssyy (green and wrinkled) will result in a single smooth yellow phenotype (SsYy). That's the goal many breeders aim for, reducing phenotype variation. To create a true bred homozygous "strain" that will also produce offspring with the same reduce phenotype variation (generation after generation) the genotype variation also needs to be reduced, which limits the amount of suitable plants to those that are homozygous (so you never get the Ss or Yy combination but only SS or ss and YY or yy).
As shown in the F1 above for example, if you'd select and cross two SsYy plants (smooth round yellow) from the F2 generation you'd end up with exactly the same variation in F3 as you had in F2, including some green and some wrinkled again.
Obviously with more relevant traits in the mix the variation increases exponentially and obviously for some breeders it will take longer than others (luck factor playing a larger role too). You can pick smooth yellow for 10 generations and still produce green wrinkled in the offspring (as long as there's s and y in the mix they will be recombined in some offspring).
Anyway, point is not at all that F3 is IBL or stable already, point is F6 or F7 may be considered IBL, it does not automatically mean it's homozygous as in a stable IBL.
The F2 will have more variation than F1 (to create a new strain you create variation first) which you start to stabilize in F3, and will take several more generation before you end up with a stable iBL, and requires growing out a lot of plants.
