# Aion model not working on my M3 Ultra 96 GB running macOS 27

**URL:** <https://community.topazlabs.com/t/aion-model-not-working-on-my-m3-ultra-96-gb-running-macos-27/105142>\
**Category:** Bugs and Issues\
**Tags:** investigating\
**Created:** [September 22, 2026, 10:50pm UTC](https://community.topazlabs.com/t/aion-model-not-working-on-my-m3-ultra-96-gb-running-macos-27/105142 "2026-09-22T22:50:29Z")\
**Posts on this page:** 4\
**Page:** 1

<div class="post-metadata">

**Author:** ![Chamberhouse](https://community.topazlabs.com/user_avatar/community.topazlabs.com/chamberhouse/32/241077_2.png) [@Chamberhouse](https://community.topazlabs.com/u/Chamberhouse)\
**Post date:** [September 22, 2026, 10:50pm UTC](https://community.topazlabs.com/t/aion-model-not-working-on-my-m3-ultra-96-gb-running-macos-27/105142/1 "2026-09-22T22:50:29Z")

</div>

The Aion 16x slo-mo model doesn’t work on my M3 Ultra with macOS 27. However, it works on all other combinations I have tried - i.e. previous macOS versions on all machines and macOS 27 on my Studio M1 Max and M2 Air.

Also, it’s the model itself, not the app version. The same problem occurs with Aion in TVAI v7.

There is no error reported, it just stays on “Estimating” indefinitely in the export queue. I have noticed that almost all available memory is used as well (around 90 of 96 GB), until I stop via “Close video”.

Fortunately, I have dual boot with macOS 26 on my M3 Ultra, so it’s not an urgent issue for me.

I noticed that according to a recent benchmark, Aion on an M5 Max 48 GB / macOS 27 works OK.

Anyone with an M1 / M2 ultra on macOS 27 having a problem with Aion?

Andy

---

<div class="post-metadata">

**Author:** ![Ren](https://community.topazlabs.com/letter_avatar_proxy/v4/letter/r/ed655f/32.png) [@Ren](https://community.topazlabs.com/u/Ren)\
**Post date:** [September 23, 2026, 3:11pm UTC](https://community.topazlabs.com/t/aion-model-not-working-on-my-m3-ultra-96-gb-running-macos-27/105142/2 "2026-09-23T15:11:23Z")

</div>

Same issue with my M2 Ultra. I can’t even run the built in benchmark as it stops when it gets to the Aion portion.

---

<div class="post-metadata">

**Author:** ![Chamberhouse](https://community.topazlabs.com/user_avatar/community.topazlabs.com/chamberhouse/32/241077_2.png) [@Chamberhouse](https://community.topazlabs.com/u/Chamberhouse)\
**Post date:** [September 23, 2026, 4:24pm UTC](https://community.topazlabs.com/t/aion-model-not-working-on-my-m3-ultra-96-gb-running-macos-27/105142/3 "2026-09-23T16:24:42Z")

</div>

It was via the benchmark that I discovered the problem, as I seldom use Aion. In the meantime, you could remove Aion from the benchmarks by deleting the following from the benchmarks.json file:

```auto
  "16X Slowmo": {
    "Aion": {
      "type": "FrameInterpolation",
      "factor": 16,
      "model": "aion-1"
    }
  },

```

That json file is in:

```auto
'/Applications/Topaz Video.app/Contents/Resources/models'

```

Thanks.

Andy

---

<div class="post-metadata">

**Author:** ![DocZed](https://community.topazlabs.com/letter_avatar_proxy/v4/letter/d/ecae2f/32.png) [@DocZed](https://community.topazlabs.com/u/DocZed)\
**Post date:** [September 24, 2026, 4:45am UTC](https://community.topazlabs.com/t/aion-model-not-working-on-my-m3-ultra-96-gb-running-macos-27/105142/4 "2026-09-24T04:45:38Z")

</div>

The benchmarked M5 Max is mine. On 27 all models start, but none can finish – except when benchmarking. According to release notes 1.7.1 is currently incompatible.