Route Drafts Cheap Before Pro Credits Touch Hero Shots

Design and content desks burn budgets in a predictable way: every sketch runs on the most expensive model because “quality” feels safer. The ledger disagrees. Rough composition checks do not need hero-shot spend. Final posters do. If your image generator stack cannot separate those stages, you will empty a monthly credit pool on drafts that never ship — then pause generation until the next refresh while the real campaign still waits.

This guide treats model choice as a routing problem, not a brand loyalty test. AI Imagine publishes a clear floor on generation cost — Standard model runs start at 2 credits, Pro at 6 — and documents Nano Banana tiers with different speed and finish jobs. The pass rule is simple: drafts stay cheap; only keepers climb. The rest of the workflow below assumes you will actually read the cost line inside AI Imagine before every run.

Why Pro-First Drafting Quietly Empties The Month

A font or brand designer mocking posters will often test five layouts before one composition locks. If each test burns Pro credits, the month dies on exploration. Pricing is explicit that unused subscription credits do not roll over — they expire at the end of the billing month — so overpaying early is not “investment.” It is waste you cannot recover next cycle.

The occupational fear is concrete: the campaign still needs hero frames, but generation pauses because the draft week already spent the pool. That is a capacity discard on the credit ledger, not an aesthetics debate. Count discarded sketches that never left the draft folder; if most of them ran at Pro cost, the routing failed before the art direction did.

Read The Credit Floor Before You Touch Style Prompts

AI Imagine’s pricing page frames subscriptions around monthly credits with all models and resolutions unlocked on paid tiers. Yearly billing saves 33% (pay for 8 months, get 12 months of credits issued monthly). Credit packs, when you buy them, stay valid for one year — unlike monthly subscription leftovers that expire.

Plan	Monthly price	Credits / month	Rough image ceiling (published)
Starter	$4.99	100	Up to 33 images or 10 videos
Standard	$9.99	300	Up to 100 images or 30 videos
Premium	$29.99	1,200	Up to 400 images or 120 videos
Studio	$79.99	3,600	Up to 1,200 images or 360 videos

When monthly credits hit zero, image generation pauses until refresh unless you upgrade or buy a pack. That pause is the fail signal to design around — not a surprise to discover on launch day. Build the week’s ladder against 2-credit Standard probes and 6-credit Pro finishes before anyone argues about style words or campaign adjectives.

Split Nano Banana Stages Instead Of One Expensive Default

The site’s own Nano Banana comparison is useful as a routing map: Standard for trying ideas (~2s), Nano Banana 2 for daily images (roughly 4–6s, high quality, lower cost than Pro), and Pro for important visuals (slower, best detail, highest cost). Resolutions in that comparison go up to 4K on Nano Banana 2 and Pro, with Standard listed at 1K. Do not treat that as a taste lecture. Treat it as a spend ladder.

Standard For Layout And Prompt Direction

Use Standard when you are still hunting composition, crop, and whether the idea even works. The job is rejection speed. If a layout fails, you want that fail cheap. Burning Pro here is how months die early.

Nano Banana 2 For Daily Keepers

Once the layout direction is stable, move to Nano Banana 2 for most shippable stills — site imagery, social graphics, routine ecommerce frames. The published note that NB2 is more stable for text-in-image (including Chinese characters) matters when mockups include words; for poster and packaging tests, that stability is part of the keep/kill decision, not a novelty.

Pro Only For Hero And Paid Placements

Climb to Pro when the asset is a homepage hero, ad creative, or other high-stakes frame where detail and lighting finish justify 6-credit runs. If you cannot name why this frame is a hero shot for a real placement, it is not a Pro job yet.

Workflow Steps Inside The Generator Credit Ladder

Open the Image Generator inside AI Imagine, choose Text to Image or Image to Image, set Provider/Model, write the Prompt, and read the cost line before you generate. The run panel shows remaining credits beside the cost — use that as a cockpit, not a decoration. In my testing protocol, the ladder is fixed before the first click: Standard probes first, Nano Banana 2 second, Pro last. Skipping the order invalidates the spend read.

Step One Lock One Source Still

Keep one source still fixed across the ladder so you are not comparing unrelated scenes. Write the exclusions into the prompt: no invented logos, no unreadable type, no second product that was never in the brief. If type comes back unreadable, that frame is a hard stop — discard it before any credit climb.

Step Two Climb Only After A Keeper

Run three Standard probes for direction. Promote one candidate to Nano Banana 2. Open Pro only if that NB2 keeper still fails a detail check against the brief — warped type, muddy edges, or missing product parts. If the miss is conceptual, change the brief; do not “fix” a bad idea with a more expensive model. A week that burns Pro credits on sketches that never shipped is a routing failure, even if one frame looked impressive in isolation.

Observable pass rules:

1. Draft stage never opens on Pro by default.
   
   
2. Any invented logo or unreadable type fails before spend climbs.
   
   
3. Hero spend happens only after a keeper already exists at NB2.
   
   
4. If the month pauses mid-campaign, the routing failed — fix the ladder before buying panic packs.

Track the credit ledger in plain numbers: Standard probes at 2 credits each, Pro finishes at 6. If draft-stage spend exceeds hero-stage spend for two full cycles, the desk is still climbing too early and needs a written stop rule.

Where This Routing Still Needs A Human Gate

Routing does not replace brand review, legal checks on likeness, or a human eye on type hierarchy. It also does not fix a chaotic brief that changes every stand-up. If inputs change every ten minutes, cheaper drafts only produce cheaper chaos. Keep Pro reserved only for frames that already survived a careful human skim and a written placement name on the brief.

Climb Spend Only After A Keeper Exists

If you need one pretty experiment, any default model will entertain you. If you need a month of posters and product stills, route generations by stage: Standard to find direction, Nano Banana 2 for daily keepers, Pro for heroes. Watch the 2-versus-6 credit floor and the no-rollover month inside AI Imagine. The desk that climbs only after a keeper exists will outlast the desk that treats every early sketch like a launch-ready campaign asset. Keep the ladder written where the team can see it, or the old Pro-first habit returns under deadline pressure without anyone noticing the burn.