Pricing.

Pay-per-call. No subscription. No minimum.
You pay for what you generate, billed via your API key.

42 Image Creative
$0.20 per image (any format)

1080×1080 · 1200×628 · 1080×1920

See the API →
42 Video Creative
$1.50 per finished ad

4–8s footage + motion graphics + CTA

See the API →
42 3D Creative
tiered per generation

Lite $0.08 · Standard $0.15 · High $0.25 · Ultra $1.50

See the API →
AI Data Transformer Beta
$0.05 per call (BYOK)

Bring-your-own-key for the LLM. Pricing covers our compute only.

See the API →
42 CAD Creative Beta
TBD private beta

Pricing finalised post-beta. Request access for early-access terms.

See the API →

Subscription plans coming.

Free / Builder / Growth / Scale tiers are on the roadmap. Pay-per-call rates will not change — subscriptions will bundle quotas at a lower per-unit cost. Drop us an email and we will tell you when they launch.

Notify me →

FAQ

How does billing work?

Pay-per-call. You pay for what you generate, no monthly minimum. Charges are processed via your API key on every request.

Are there subscription plans?

Not yet. Subscription tiers (Free / Builder / Growth / Scale / Enterprise) are on the roadmap and will not change pay-per-call rates — they will offer included quotas at lower per-unit cost. Drop us an email below if you want notice when they launch.

What does BYOK mean for the Data Transformer?

You provide your own LLM API key (Groq, Anthropic, OpenAI, or Google Gemini). The $0.05 per-call fee covers our compute only — the LLM inference cost is billed by your provider directly. Free Groq tier works for prototyping.

Are there volume discounts?

Get in touch if you ship sustained volume — we discuss case-by-case. There is no published volume tier yet.

What happens if a generation fails?

It depends on the API. Image and Data Transformer charge after the generation succeeds — failed runs are not billed. The 3D Creative API charges at the start of the generation, so you pay even if a specific run fails to produce a usable mesh; this is documented in our actor source code, we want you to know up front. Successful generations that you simply do not like are not refunded — preview prompts in the public terminal at /tools first.

How accurate are the listed times for image / video / 3D?

They reflect typical ranges sampled from the public feed at /tools. Multi-paragraph prompts and peak GPU queue times can extend the long tail. The pricing is per-call, not per-second, so latency does not change what you pay.

Can I see the math behind these rates?

Yes. Our cost breakdown (provider stack: Vertex AI Imagen 4, RunPod for video and 3D, Tencent Cloud for 3D Ultra tier, Apify compute) is documented internally and we are happy to share the methodology with serious customers. Email us.