How to Optimize Azure Function Apps for Cost and Cold Start Issues?

I'm the cloud architect for a small company with around 20 people, and we have a number of algorithms and models running in Azure. Cost is a huge factor for us, but we also need our architecture to be scalable. I've been leveraging Function Apps since they can scale to zero, keeping costs down, while also being able to scale up during high usage. I've organized each algorithm into its own function to facilitate independent scaling and development. However, we've started facing issues with cold starts, which can take a few minutes, leading to a frustrating wait for users. While the actual computation takes seconds, this wait time is frustrating. I've considered using the Flex Consumption Plan to keep a single instance warm, but we can't combine multiple function apps into one, and that would drive costs up. I'm exploring my options: 1. Keeping separate function apps on a standard app service plan, which limits per-function scaling; 2. Consolidating into one function app on a Flex plan, though that would require a major refactor; or 3. Our current workaround of a warmup call when users log on, which helps a little but isn't a real fix. I'm looking for advice to balance costs, cold starts, and scalability without facing constant restructuring based on Azure's pricing changes.