Usage & costs

Every generation has a real model cost, and Hype Flow records it at the moment it’s billed — per run, per model, per provider.

You see cost before you spend, too: a graph run shows its estimated cost per node — model and total — before dispatching, so a heavy run never surprises you.

The usage dashboard

Organization admins get a usage view with:

• Daily spend — a day-by-day chart of generation costs.
• Breakdowns — which models and which kinds of work (image, video, audio, LLM) the spend went to.
• Recorded per generation — costs attach to the actual runs, so a spike always traces to the project and node that caused it.

What drives cost

• Video is the heavy tier. Duration × resolution is the multiplier that matters; a 1080p clip costs a multiple of its 720p draft.
• Batches multiply. A Collection run costs roughly per-image × count — skim the plan before large counts.
• Premium image models (the -pro / -premium tiers) cost several times the standard tiers. Draft on a standard tier, finish on premium.
• LLM steps are comparatively cheap — prompt expansion is rarely the line item that hurts.

Practical habits

• Draft at low resolution / short duration, then re-run the keeper at delivery quality.
• Use per-frame regeneration on collections instead of re-running whole batches.
• Check the daily chart after a heavy campaign day — surprises should be boring and explainable.

Limits and provider configuration are managed by your administrators.