I have iStrelok and Ballistic AE on my phone and find they differ on data about 2 MOA at 600 yards.  Is it normal for applications to differ this much?

All cartridge and environmental data are the same.

No thats pretty substantial and like there is a bad input somewhere.

Copy all.  I will rerun the data and see where I goofed.

I found they all match pretty good. Like within an inch and 1fps at 1000+ yards.

Zero distance, scope height over bore, elevation or DA vs. temp/humidy/pressure, zero atmosphere, powder temp, coriolis etc.
Some apps default to certain settings some don't, everything should by input manually and all the app settings should be gone through to look for hidden things before comparing 2 apps(and for accurate solutions).

Check you're using the same BC data base (G1 or G7) in both.
Does one have temperature compensation active?
Are the Temperature Comp setpoints the same in both?
Do you have spin drift active on one and not the other?

They should be pretty close.  I have been running Ballistic AE and TRASOL together at the last few matches, and at most they are only different by 0.1 mils.  When they are different Ballistic AE is a tenth lower than TRASOL.

IStrelok had the Temperature sensitivity set to 1.1. I zeroed that out and they match.

That fixes the difference, but what is the temperature sensitivity of your load?

I don't know and for the most part it won't matter.  I just reach to 600yds on IPSC steel every now and then and when I deer hunt I don't shoot past 250yds max.  If I was shooting for tiny groups at 600yds I would be more concerned.

When I sight in for deer season I go to 300yds and verify I am on an 8 inch plate with every shot.

I did notice the difference in the tables really started to show up at 350+.

FWIW, the load is 27.5gr of AR-Comp under an 123 ELD-M going 2450 out of my 16in Grendel.  In case you want to run the numbers.  It is just under 16MOA of drop at 600yrds when sighted in at 200yds.

In the course of development of the ballistics utility that I publish (BallisticXLR) I found that the maximum acceptable deviation (acceptable to me as a publisher of a ballistics utility) was < .5MOA (~.2mrad) @ transonic distance given identical inputs. Any more than that and I'd suspect something is wrong in the inputs or a bug in the app. Any notion of a bug in a particular app is something you should look askance at because the mathematical models are pretty well fixed even though there are a number of ways to implement them. Still there's not a lot of ways to screw the math up but there are >20 easy and sometimes subtle ways to screw up the inputs. When I was developing mine I ran thousands of comparisons between JBM, mine, Shooter, and a few others. The usual difference was just under .1mil (~.2moa) between the highest and lowest values predicted by all of the apps.

One of the big ones to look at is "click value" or how it defines a minute of angle. I've seen the weirdest things come from those and they can be hard to figure out if you don't know what you're looking at. Given the symptoms you describe and how they're coming on, I'll bet that that's your issue. As an example, in BallisticXLR I use the MOA:inch@100yrds relationship as a tunable parameter to correct for scopes with minor variations in click value. 0-4% deltas are common. What I'm doing with that is making sure that the shooter dials to the indicated angular value that the DOPE says to even if the indicated angular value is not equal to the real angular value. That kind of thing starts showing up around 300-400 and really gets pronounced by 600 with most long range capable setups. FWIW, "long range capable" should in this context be taken to mean ballistics roughly equivalent or flatter than a G1 BC .420 or greater on the projectile with a MV 2400fps or greater. That definition does not apply outside of this context.

Good info.  I did not mess with the click values.  I left it at .25 MOA.  Note: I mentioned the MOA deviance not the clicks.   Does your app allow for the temp sensitivity of the powder to be included in the variables.   I can see it as valuable if I develop a load at 90F and plan on shooting it competively in 35F temps at greater than 500yds.  I shoot for fun at steel out to 600 and the noted deviance I saw comes up to about a foot.  That can have me off the plate at that range.

I am going to run the dope this weekend and see the truth data.  iStrelok was pretty good in the past.

The published G1 for the 6.5 123gr ELD-M is .506 and my MV is 2450, just barely above you threshold.

Yes. Temperature sensitivity has been part and parcel to how BallisticXLR works since the beginning partly because air temp affects ammo performance but more because accurate air temp is crucial to good ballistics calculations and because you can slide 20deg up or down to account for changes in air pressure associated with 1000' of elevation change. The way my tables are laid out, the shooter can account for changes in air temperature, MV or elevation/pressure by just moving the requisite number of columns left or right.

I'll assume you're shooting a Grendel or .264LBC (aka a grendel). Temperature induced velocity changes are sometimes but not always severe enough to have significant effect. That's really distance dependent but usually really starts to show up after 500yrds. Air temperature on the other hand finds it very easy to have a pretty dramatic effect on POI and I see those differences a good bit earlier than 500yrds but the magnitude is small at close distances so most observers don't notice, it gets lost in the noise.

Where I shoot there's often a 50-70deg difference between early morning and mid-afternoon temperatures. At 1000yrds a 50-70deg air temperature difference makes for .5mrad difference with my extremely flat shooting 6XC while a 50fps difference makes for only a .3mrad delta.

For you it's 1-2 full mrad depending on starting temperature (so 3x as much as my 6xc) for a 50-70deg span but for a 50fps velocity difference in any one temperature regime it's about .5mrad difference. So, air temperature is more important to ballistic trajectory than even notable velocity changes BUT that is compounded by the fact that air/ammo temp also affects resulting velocity. Stack imprecision of air temperature together with imprecision in velocity and you can see very large departures from calculated to observed POI.