Live script efficiency

Hi!

I’ve been using Live Scripts in Raven to create assembly elements in Rhino. It has changed how I work, and I’d really like to keep using it, but I’ve run into a problem.

The project is small. So far I’ve only added two slabs and two wall types, placed in six locations. Even so, Raven has become much slower and uses a lot more tokens than it did at the start. A task that used to take about 600 tokens and under a minute now takes around 19,000 tokens and more than 30 minutes. At this point it’s close to unusable.

My questions:

  • Is this because I’ve gone over some usage limit, so I’m being throttled?
  • Or does Live Scripts slow down when several scripted elements are stacked in the same file? For example, does it re-read or re-evaluate all existing elements each time I add a new one?

If it’s the second, is there a recommended way to structure the file (separate files, baking elements, splitting scripts) to keep it fast?

Setup: R4V3N 2.1.18, Rhino 8 SR37, Windows 11. Model/plan: Personal

Thanks!

Hi Torgeir,

as to your questions:

  1. You are not being throttled, but maybe you switched between ai models - in general there are different ai models available. Some are premium (expensive, strong) models (eg. ChatGPT 6 Astra) and some are cheaper, faster and a little bit weaker (eg. ChatGPT 6.1 Sol or Terra). Using the premium model will use up more tokens, but often leads to better results in complex tasks.

  2. To your specific case: It’s hard to judge from just a screenshot, but in general, when the context grows, the AI models start getting more and more confused. So yes, stacking the parts of your model in one single script does make it slower. It is better to work with smaller steps in a modular way. It also helps to start a fresh chat every now and then, eg. when starting work on a new part of the model.

If you want more detailed support, send me a message to support@raven.build including your Raven Account Email and I can look into it further.

Max