I originally left a 1-star review because most of what’s included in the lifetime deal felt unusable for real work. I updated my rating to 2 stars after you added Supermachine Next Gen, because that was a meaningful improvement and I want my review to reflect progress when it happens.

Why I update my reviews (and why that’s normal)

Rex, you criticized me for making multiple edits and implied that isn’t “normal.” I strongly disagree. Updating a review when a developer fixes issues is exactly what a fair reviewer should do. Most people leave a rating and never revisit it—I do the opposite: I adjust the rating when the product changes. In fact, you missed that I raised the rating from 1 to 2 specifically because you added Next Gen and I gave you credit for it.

What’s working

Right now, only two models feel consistently usable without heavy workarounds:

Supermachine Dream

Supermachine Next Gen

These can produce decent results “out of the gate,” which matters because most customers don’t want to become prompt engineers just to get a clean image.

What’s not working (and why this is still 2 stars)

Everything else included in the lifetime deal is still not competitive in quality. The recurring issues include:

frequent anatomy/hand errors (the “Freddy Krueger hands” problem)

incorrect digits / morphing

muted or flat color

a “soulless” / overly artificial look

requiring excessive negative prompting just to avoid basic mistakes

For a $400 lifetime deal, the gap between what’s promised and what’s consistently usable is still too large.

About “promoting other companies”

I’m not trying to promote anyone. When I reference other tools or what’s possible elsewhere, it’s not advertising—it’s benchmarking. Customers compare products. That’s normal. The point is to show what current quality and update cadence looks like in today’s market, and what users reasonably expect from a paid platform.

The core request: add modern open-source models

I understand the argument that you can’t include every expensive proprietary model forever—that’s fair. But what I don’t understand is why you haven’t added strong open-source, permissively licensed models (for example, Apache 2.0-licensed options such as Qwen Image Edit, HiDream, Dream Omni 2, and Z-Wave, among others).

These models (and their optimized/quantized variants) are already usable on consumer GPUs. If they can run locally for many users, it’s reasonable for customers to ask why a paid platform isn’t keeping pace—especially when the lifetime deal is positioned as long-term value.

What I want to see

If you want higher ratings and better retention from LTD buyers, here’s the path:

add stronger modern models (especially permissive open-source options)

improve default quality so results don’t require heavy negative prompting

publish a clear roadmap for model updates

actively incorporate recurring customer feedback instead of dismissing it

Bottom line

I updated to 2 stars because Next Gen was a real step forward, and I gave you credit for that. I will not move this to 3–5 stars unless the overall lifetime offering improves substantially and the model lineup becomes consistently usable beyond just two models.

If you implement this feedback and meaningfully modernize the lifetime deal, I’ll revisit the rating again—upward—because that’s how I review: based on improvements and results.