LLM Observability with AI SDK with images

Feature request

Is your feature request related to a problem?

It is not related to a problem

Please describe.
Currently I am using withTracing wrapping my image models and I am absolutely loving it because it helps me track how much each model, user, and chat is using which has helped me optimize multiple issues. However, when I tried implementing it into the image models I got an error (This might also be the case with the new stuff Vercel AI SDK is adding like transcription and voice).

Describe the solution you'd like

Let the withTracing take in Image models (and potentially other types of models).

Describe alternatives you've considered

I have considered tracking it manually with the finishing function that lets you see the tokens used, cost, etc. However, I feel like it would be a more natural solution and better for the future if this was added.

Additional context

I might not have the posthog newest version so sorry if this is already done.

Current posthog versions:

    "@posthog/ai": "^4.4.0",
    "posthog-js": "^1.245.1",
    "posthog-node": "^4.17.1",

Debug info

- [ ] PostHog Cloud, Debug information: [please copy/paste from https://us.posthog.com/settings/project-details#variables]
- [ ] PostHog Hobby self-hosted with `docker compose`, version/commit: [please provide]
- [ ] PostHog self-hosted with Kubernetes (deprecated, see [`Sunsetting Kubernetes support`](https://posthog.com/blog/sunsetting-helm-support-posthog)), version/commit: [please provide]

Start by reading the @posthog/ai withTracing integration and the Vercel AI SDK image-model path mentioned in the issue. Reproduce the reported error with the listed package versions, then determine the existing model-type boundaries. Done means image models can be wrapped by withTracing without the reported error, with the supported scope documented.