In Unearthed Arcana: That Old Black Magic (2015), they playtested a Tiefling Variant with an Abyssal Tiefling subrace. The Abyssal Tiefling gets this Abyssal Arcana trait instead of the original Infernal Legacy trait:
Abyssal Arcana. Each time you finish a long rest, you gain the ability to cast cantrips and spells randomly determined from a short
list. At 1st level, you can cast a cantrip. When you reach 3rd level,
you can also cast a 1st-level spell. At 5th level, you can cast a
You can cast a spell gained from this trait only once until you
complete your next long rest. You can cast a cantrip gained from this
trait at will, as normal. For 1st-level spells whose effect changes if
cast using a spell slot of 2nd level or higher, you cast the spell as
if using a 2nd-level slot. Spells of 2nd level are cast as if using a
At the end of each long rest, you lose the cantrips and spells previously granted by this feature, even if you did not cast them. You replace those cantrips and spells by rolling for new ones on the Abyssal Arcana Spells table. Roll separately for each cantrip and spell. If you roll the same spell or cantrip you gained at the end of your previous long rest, roll again until you get a different result.
Abyssal Arcana Spells
||Spare the dying
||Tasha’s hideous laughter
I would assume that homebrew could trust this list to be at least somewhat balanced as possible spell replacements, since it made it to UA. However, keep in mind that UA is not strictly canon, but rather playtest material from WotC game designers.
In that regard, the Infernal Legacy trait looks a lot like a Dragonmark (from Unearthed Arcana: Eberron (v1.1)) or the Magic Initiate feat (PHB, p. 168). You could also skin one of those feats into a racial ability, as a homebrew that wouldn't deviate too much from the rules.
And completely RAW, you could always do a Feral Tiefling with the Devil's Tongue trait (SCAG, p. 118), too.
XGtE is optional
It's important to remember that most of XGtE's content comprises optional rules, and this particular section on character generation is no exception. In fact, the opening on page 61 explicitly states:
IDEAS, NOT RULES
Even though these pages are full of tables and die rolls, they don’t make up a rules system — in fact, the opposite is true. You can use as much or as little of this material as you desire, and you can make decisions in any order you want.
Stretching that a little, we could infer it to mean that despite what might be implied by the options presented here, none of it should be taken as informing the actual rules of the game. It definitely doesn't help that this section of Xanathar's does contradict the racial history given for tieflings in their own description in the PHB. But if we want to take the content as presented and try and make it work, we could consider the variability of genetic expression.
All cambions are half-fiends (but not all half-fiends are cambions)
Simply put, though a cambion is a half-fiend and is always the result of union between a fiend and a humanoid, the union of a fiend and a humanoid does not always result in a cambion. Sometimes, when a devil and a humanoid procreate, perhaps the result looks more like a tiefling.
To draw a crude analogy to the real world, children of mixed race parents can vary wildly between strongly expressing the racial characteristics of one or the other parent or appearing somewhere in between - depending on exactly what random bits of genetics get passed on and in which combination. It's not even unusual for children from the same parents to appear to be completely different races!
Obviously the genetics of a fantasy world are much more complicated than ours could conceivably be (especially when you allow for magical factors). It is not much of a stretch to imagine that the children of a devil/humanoid pairing may sometimes express much more of their fiendish heritage (producing a cambion) and sometimes less (producing a tiefling).
What's the difference?
The general difference between a tiefling and a cambion is that a cambion is a very strong expression of fiendish ancestry that perhaps you can only get with one directly fiendish parent, but a tiefling is a much lesser expression of such ancestry which can also occur in descendants far removed from the original fiendish influence.
Almost human, with a variety of fiendish traits.
Before D&D 4th edition, tieflings were simply the descendents of humans and evil extraplanar beings, such as demons or devils. 4e retconned them to be humans who had acquired devil-like features in a pact for power with Asmodeus.
When tieflings were introduced in AD&D 2nd edition, they were originally humans with a mysterious fraction of otherplanar ancestry. This was implied to be that of a demon or devil, but it was not explicitly stated, and most tieflings were wanderers and orphans who did not know their own ancestry. The Planescape Campaign Setting (1994) describes them thus:
The Planewalker's Handbook (1996) expounded on their origins, and gave the tieflings a variety of traits:
Examples of traits include fangs, small horns, pointed teeth, unusually colored hair, eyes or skin; a tail, goatlike legs, small feathers, spiny ridges, or fur. In other words, they were humans who had some feature normally associated with their ancestor, who was a demon, devil or similar evil outsider.
Tieflings in 3e were described in the 3.5 Monster Manual (p.210) as appearing very human:
In other words, tieflings prior to being "claimed" by Asmodeus generally looked human, enough so to pass for human at a glance, but closer examination would reveal some trait like small horns or red eyes. Many looked sufficiently human to even pass as human.