We have a new project with a new customer.
We start know with development. The commissioning and go live should be in Janyary 2026.
We Use Postgres as Datasource, AD, Python Scripts, OPC UA to connect to Devices, Perspective, Gateway Redundancy.
My Question is, should we start with 8.1 and migrate next year to 8.3? Or it is the better way to start with 8.3.
Thanks
The release train is ~5weeks, so we’ll be around .3 which is still quite young in terms of fixes/patches. I would go with 8.1 for Jan 2026 with a definitive plan to migrate to 8.3 once thoroughly tested/compared against the 8.1 production version.
The biggest advantage I see for a new install is the integrated historian with QuestDB and the Perspective drawing tools. If you plan to use Postgres and don’t have issues currently developing in Perspective, I would personally wait until 8.3 has some more time under its belt before upgrading. Especially if this is a new customer's first exposure to Ignition. You want the install to be a huge success, and give the customer a positive experience with Ignition.
8.3 with no doubt
You should definitely start with 8.1 and then migrate later.
I agree. Start with 8.1 and also read the upgrade advice docs that will be coming shortly so you can avoid anything that may not be well supported in 8.3 verses 8.1. I know there were a lot of gotchas with 7.9 to 8.0 upgrades.
During the Dev panel discussion at ICC yesterday a similar question was asked. It was something like when will 8.3 be considered stable or LTS (something like that) and they said that they were confident in what was released to be stable and is already considered to be LTS because it is version 8.3 in which odd numbers are considered LTS versions.
IMO, there’s a reason most software companies alternate between LTS and non-LTS versions 
If every version is LTS, no version is LTS.
The problem with our current LTS and versioning strategy is that LTS is only an indicator of how long the version will be supported and not necessarily an indicator of stability, though it's supposed to (eventually) indicate stability as well.
As the new versioning and LTS scheme is implemented over the next few years I think this will change, and LTS releases will also be truly stable from basically the day they are cut and released.
For your timeframe, I would use v8.3 immediately. v8.3 is unlike any past Ignition major release--it is very solid right out of the gate.
Do check for 3rd party modules that aren't yet ready. (Mine are still betas for a couple more weeks, I think.)
I’m with Phil on this one, everything I’ve seen suggests that the QA process on this release has been first class. The software is 100% able to be developed on, especially if you deploy in January, there’s two versions due that will fix minor bugs during your development cycle.
8.3 for 2 reasons, excluding new features:
I would not present this as a selling point 
Better to find and report the bugs now than on the day you go live, isn’t it ?
Well, I doubt most developers start a project with the goal of finding and reporting Ignition bugs, hoping they’ll be fixed before production. That was my point.
I understand, but 8.3 is such a big step up from 8.1 that it would be a shame not to use it.
And if you’re gonna use it, you might as well do it now so you have time to iron things out, instead of updating in a few months when you’re already live.
From a customers standpoint, who is only interested in performance and uptime, I don't know if the risk is worth the reward for an immediate rollout of a version in its infancy. The deployment features, drawing tools for perspective, Gateway UI, etc are all more for an Integrators benefit from what I see. I’m more involved with traditional factory HMI control with Vision and a couple of gateways (MES and Visualization). Widget makers are risk averse, hence why there are daily posts with people using 7.9. Based on the OPs customer requirement, how would you present to the customer the cost benefit of developing in 8.3 now, vs 8.1? I sincerely ask, because 24 hours of downtime in my customers factory is $500K of lost revenue, and I'll be faced with this decision as well mid next year.
We actually asked the IA folks when we should be moving up to 8.3 during a meeting we had with them a few days ago, and their answer was to wait a couple months, make sure things are stable and all.
BUT, they also made it clear that quality control for 8.3 is higher than it’s ever been before, and we were talking about projects that are already in production.
A brand new project will not be ready before the couple months we were advised to wait.
Starting development in 8.1 now means having to move to 8.3 when everything is already live (or keep running an old version forever).
Starting with 8.3 means using the new API instead of functions that are now deprecated, having access to version control and modern pipelines (which is an improvement even from the customer perspective: it’s safer and faster).
Perspective’s performances should also be better, the secrets mechanism and deployment modes will make IT happy…
Maybe there’s a small risk, but so does migrating to 8.3 later, when things are already running in prod. I’d rather avoid the migration process on a live instance.
Your situation is different: you’re talking about updating something that’s already live. This does require regression testing, making sure that nothing breaks, etc.
For a new project, there’s really no reason not to jump straight to the latest version.
I mean, if 8.3 wasn’t a thing, would you start dev with 8.1.24 because the latest 8.1 version is too recent ?
Since this is a new project, not an upgrade, I'd argue that avoiding a future upgrade when v8.1 LTS ends is an entirely worthy uptime win.
Precisely why the OP (an integrator) should use it.
Anyways, the OP is not upgrading. It is a new project that deploys several months from now. I think using 8.3 for it is a slam dunk.
The 8.1.24 to 8.1.48 comparison isn’t valid. That’s a fairly minor feature set of improvements and bug resolutions, not a fundamental architectural engine change.
Living in the PLC world, especially Rockwell, maybe it’s an unfair comparison, but new versions of rockwell, even supposed minor, breaks sh** all the time. Rule of thumb, stay a year behind.
I do agree with your point that eventually there is risk, whether starting on 8.3 or upgrading later. However, just like with 8.1, later should still offer a more vetted version than now, so as time goes, the risk diminishes.
End of day, lots of good arguments both ways, and OP and their customer have to weigh the risk/reward.