I always thought the only things exclusive to the Pro license was to be able to use multi-GPU and to use Topaz commercially. Now we’re locking down certain models to the Pro license as well?
I’m not happy about this…I don’t think you should lock usage of newer models to the Pro license only. Some of us don’t care to use the cloud for rendering.
Select the model then press “CTRL + E”
That will start the local export.
Interested to test this one.
and the fps label keeps switching between numbers that makes sense and this:
Thank you. I didn’t know you had to scroll down to complete the process of downloading the model.
So even an RTX-4070 will not work to try SLS 
Hi, I’m the head of research here. More opinions about quality of the model are welcome. Starlight, Starlight-Sharp and Starlight-mini are the same suite of models we developed last year. 100% guarantee that it’s much better than any open source model and we own the whole IP. If there is any chance in the future we utilize an open source model to serve our customer, we will let our customers explicitly know it in the first place.
I would but TL does only serve cloud credits (i dont know what i should do with it, i don’t use the cloud), two years ago we got license extensions, that was a way more attractive.
And my licence is out of the beta time frame.
I’m sure that would be very much appreciated, thank you.
You will always be PRO in my heart 
I think Starlight Sharp is so fast because only 4 images are used for calculation instead of 8 or 16. Unfortunately, after every 4 images, I see a slight jerkiness in the patterns in the image and unfortunately, slight fluctuations in brightness. I think Starlight Mini is better at the moment. I hope the speed optimizations in Starlight Mini are without any loss of quality!
Dear Topaz Labs Team and Leadership 
As a long-time user and admirer of Topaz Video AI, I’ve seen firsthand how you’ve pioneered the field of AI-driven video upscaling, image restoration, and enhancement. Your tools have set the standard for quality and usability, transforming blurry footage into stunning visuals that professionals and hobbyists alike rely on. However, in this rapidly evolving AI landscape, I believe it’s time for a bold strategic pivot to maintain that leadership—and I urge the CEO and team to consider this feedback seriously, as the competition is closing in fast.
The core issue is the pace of innovation on new AI models. While Topaz has made strides with releases like Video AI 7.1 and the recent beta introducing Starlight Sharp, the focus often seems skewed toward iterative UI redesigns with each update. These cosmetic changes—reshuffling buttons, tweaking interfaces—consume valuable development time that could be better allocated to core advancements. Users like me crave groundbreaking models that push boundaries in quality, speed, and versatility, not just a fresh coat of paint on the app.
Look at the competition heating up in 2025: Google has rolled out Veo 3, a powerful AI video generation and enhancement tool that’s integrating seamlessly into their ecosystem, offering real-time upscaling and creative features that rival or even surpass current offerings. Alibaba’s Wan 2.2 and Wan2.1-VACE models are open-source powerhouses, enabling video creation, editing, and enhancement at scale—democratizing high-quality tools and attracting developers worldwide. Other players like HitPaw, AVCLabs, and Nero AI are aggressively releasing alternatives with comparable or superior upscaling capabilities, often at lower costs or with more flexible pricing. If Topaz doesn’t accelerate, these giants could replicate (or exceed) your tech, eroding your market share overnight.
If I were the CEO of Topaz Labs, I’d immediately redistribute resources: freeze UI overhauls for the next 12-18 months and redirect that engineering talent toward rapid R&D on next-gen models. Prioritize maximum user value—make advanced features accessible to all, not gated behind “pro” tiers that fragment the community. Roll out frequent, model-focused updates with beta testing for early adopters, and collaborate with users on open feedback loops to iterate faster. This isn’t just about survival; it’s about reclaiming the pioneer spirit that made Topaz great and delivering the absolute best quality to keep us loyal.
Topaz has the talent and foundation to lead—let’s not squander it. What do you think, team? I’d love to hear your thoughts on shifting gears.
Couldn’t agree more, thank you for your post.
I have been using Video AI since v2.6.4. Here is what I see:
I started looking at the options you listed. I still think Topaz has the opportunity to differentiate themselves with quality models, local processing and appropriate pricing for enthusiasts (I am not a pro)
The Starlight Sharp model is much faster, but it is prone to memory errors that can cause the program to crash or even force the computer to restart automatically.
Anything local locked behind Pro is questionable business-wise and ethics-wise, including the multi GPU processing.
What you SHOULD be doing is having standard licensing schemas common to intangible goods like software.
Fully functional version but without cloud for the average Joe. The personal license. They can use it for their own projects but not commercially. They can upscale their star trek series in 45 years with their two GPUs at 0.1 fps, but if they are not selling the result, so be it. They can upscale their personal videos in a few weeks and so be it. They make no money from that.
Pro license that unlocks cloud rendering for people who want to use the software professionally. You want to upscale old wedding videos for money? Buy the professional license so that you can, and unlock cloud rendering with credits as you have them now.
Corporate license for people who want to use the software in large companies with many seats.
This is where the bulk of money is being made.
Trying to squeeze every last drop of money from the average Joe, will not work.
They will at best leave (and you will lose a now almost guaranteed annual subscription for the updates) and at worst they will pirate the software. What do you think will happen once a popular piece of software becomes unaffordable or simply unreasonably priced or has an obnoxious licensing?
These things have happened again and again since times immemorial, there is no need to reinvent the wheel here.
It is not enough to have high IQ to be successful, you also need high EQ among other things.
So read the room. It does not matter if you have the best product in the world. If you are unreasonable as to how you will get the funds for its development and for making a living you will end up losing the pie and the dog.
And the thing is that you do not have the best product in the world. You have something that is constantly a work in progress, and we are all constantly beta testers, whether we beta test or not. Basically, I have been paying you for years to be your beta tester… you have achieved that.
And I would not mind keep doing it if you keep things reasonable.
FULL functionality for personal licenses, but only local and with that license you cannot use the software in a commercial manner. End of story. That’s it.
Pro license for professionals or enterprise license for large corporations that DO want to use it for commercial purposes. That’s it.
The only point I disagree with is cloud rendering. There’s no point in restricting clould access to Pro when they’re charging for the credits. An amateur’s money is just as good as a pro’s.
Beta only right now @Tommy
Everyone should be able to test it, seems that there’s a bug preventing access for some Beta users
Not sure if I’m misreading your post but - Cloud access to the Starlight Sharp model should be available for all users, to be clear.
Hello, happy to compare results with open source models but this claim you made isn’t correct. All the Topaz upscaling models are built custom by our team.
Like Kevin mentioned, we will let the community know if we do anything otherwise and it will be made clear in the UI if it is an open source AI model.
