Per-Generation Pricing: Examples & Companies

4 companies in the corpus Updated stub analysis
Definition

Per-Generation Pricing is a billing unit where each AI-generated creative asset — image, video, or design — is counted as a 'generation' and metered accordingly.

Also known as: Generation-Based PricingPer-Creative-Generation Fee

What is it

Per-generation pricing is a billing unit where each AI-generated creative asset — image, video, or design — is counted as a “generation” and metered accordingly.

The generation is the natural billing atom for creative AI because users think in outputs, not compute cycles. When a marketer asks for “100 product images,” they are counting generations. FLORA, Frase, and Typeface all meter generations because it aligns with how customers measure the work done — the campaign, the batch, the content run. Choosing this unit is a value-metric decision as much as a billing one.

Platforms wire the same atom up very differently. FLORA rejects abstract credits: every generation bills at the underlying model’s published API rate in real dollars, so users see the exact cost per generation and agencies can bill clients directly from FLORA’s usage history. Frase caps “AI generations” as one of seven per-plan meters, bundling a fixed count into each tier.

The unit clusters in creative and marketing AI because outputs are countable discrete artifacts — images and short videos — unlike text, where volume is a continuous token dimension. OctoAI made the boundary explicit before its 2024 shutdown: it billed text per 1M tokens but metered its image (“Media Gen”) endpoints per image generated and per GPU compute-second.

One "generation" · four ways to package the same atom
Same generation — priced in dollars, caps, compute, or a quote REAL DOLLARS $0.151 FLORA · /image Nano Banana 2 API rate per-seat $ pool CAPPED COUNT 350 Frase · Scale plan gens/mo in $299 tier overage opt-in IMG + COMPUTE /img + /sec OctoAI · discontinued SDXL per image + GPU-sec $10 free credit BLENDED QUOTE ~$49+ Typeface · /seat/mo reported, sales-led + gen/credit usage ← EXACT DOLLAR PER GENERATION OPAQUE / QUOTED →

How it works

Each generation is a completed AI creative output. The platform increments a counter once the output is delivered, then either debits an included allowance or bills the generation at a per-unit or per-model rate. The four companies in the corpus package that same atom in four distinct ways:

CompanyGeneration unitHow it’s priced
FLORAOne image or one videoBilled at the published model API rate in real dollars (e.g. Nano Banana 2 at $0.151/image); drawn from a per-seat usage pool ($12 / $50 / $100/seat on Starter / Pro / Max)
FraseOne “AI generation” of contentCapped per plan — 25 / 100 / 350 generations on Starter $49 / Professional $129 / Scale $299 per month; overage opt-in pay-as-you-go
OctoAI (discontinued)One image output (SDXL, SD 1.5, SVD)Usage-metered per image generated and/or per GPU compute-second; $10 free signup credit
TypefaceOne brand content assetBlended into enterprise quotes: per-seat (reported ~$49–$99/user/mo) plus generation/credit usage and brand workspaces

The core unit math for a plan-capped model like Frase:

Monthly cost = tier fee + (max(0, generations_used − included_allowance) × overage_rate)

FLORA’s dollar-denominated variant replaces the abstract allowance with a real-dollar pool:

Usage spend = Σ (generations × published_model_rate), drawn down against seats × included pool; anything above the pool is billed pay-as-you-go at the same API rates.

Worked example on FLORA: a 4-seat Pro team ($50/seat/mo = $200 platform) includes a pooled usage budget of 4 × $50 = $200/month. At the Nano Banana 2 rate of $0.151/image, that pool covers roughly 1,300 image generations before overage — after which each additional image still bills at exactly $0.151, with no credit-conversion markup. That transparency is the whole pitch, and the same reasoning behind broader usage-based pricing models: meter the thing the customer already counts.


Companies using this

Four corpus companies meter AI creative outputs as generations: FLORA with dollar-transparent, model-rate generations across images and video; Frase capping AI generations inside its content-platform tiers; OctoAI (now discontinued) metering image generation per output and compute-second; and Typeface folding generation/credit usage into sales-led enterprise quotes.


Company Product Pricing modelBilling unitsFree tier Verified
FLORAAI-powered creative canvas and workflow platformYes2026-07-23
FraseAgentic SEO and GEO platform that researches, writes, optimizes, and tracks AI-search visibility for content teams.No2026-08-04
OctoAIGenerative AI inference platform (acquired by NVIDIA, sunset Oct 2024)No2026-07-30
TypefaceArc enterprise marketing AI platformNo2026-06-16

Explore this theme in the knowledge graph

FAQ

What is per-generation pricing in AI creative platforms?

Per-generation pricing meters each AI-generated output — an image, a video, or a design variant — as a discrete unit. FLORA bills every generation at its underlying model's published API rate (e.g. Nano Banana 2 at $0.151/image) so users see the exact dollar cost per generation; Frase caps 'AI generations' per plan (25 / 100 / 350 on Starter / Professional / Scale); Typeface blends generation/credit usage into enterprise quotes. The generation unit abstracts over model complexity and output format so the platform can change infrastructure without repricing the customer-facing unit.

How does per-generation pricing compare to token-based pricing for AI creative tools?

Token-based pricing is the standard for language models but maps poorly to visual content, where the 'size' of an output is pixels or frames, not tokens. Per-generation pricing uses the finished asset as the unit — one image is one generation regardless of resolution. OctoAI showed the split directly: it billed text per 1M tokens (Llama 3 8B at $0.15/1M) but metered its Media Gen endpoints (SDXL, SD 1.5) per image generated and per GPU compute-second, because token counting doesn't describe a picture.

How much does per-generation AI pricing cost?

It varies by how the generation is packaged. FLORA's paid plans start at $18/seat/mo (Starter) up to $200/seat/mo (Max), each including a dollar-denominated usage pool ($12 / $50 / $100 per seat) spent at published model rates. Frase bundles a fixed number of AI generations per plan (25 / 100 / 350) inside its $49 / $129 / $299 per-month tiers. Typeface is sales-led with no public rate card — third parties report indicative seat tiers around $49 and $99 per user/month plus generation/credit usage.

Is per-generation pricing the same as credit-based pricing?

They overlap but differ. In pure credit systems, credits are an abstract currency and a generation may cost a variable number of credits by model. Per-generation pricing keeps the generation itself as the visible unit. FLORA deliberately rejects abstract credits — it prices every generation in real dollars at published API rates so agencies can bill clients from FLORA's usage history — while Typeface and OctoAI blend generations with credits and compute.

Related billing units

Back to companies