Pricing
Prepaid credits. Pay-as-you-go, as low as 1.7¢ per image at Volume. No subscriptions, no auto-renew.
= 2.6¢ / image · 13.0¢ / video
1,500 credits · 1 credit per image, 5 per video
38% offFirst integration, prototype, weekend project.
- All 3 models · 6 resolution presets
- HMAC-signed webhooks · 5 retries with exp backoff
- Idempotency-Key · auto-refund on failed gens
- Permanent CDN-hosted output (no TTL)
- OpenAPI 3.1 · Postman · JS SDK
- Unlimited per-environment API keys
- 60 / min rate limit · credits valid 12 months
= 2.2¢ / image · 11.0¢ / video
4,500 credits · 1 credit per image, 5 per video
45% offThe sweet spot for shipping a real product.
- Everything in Starter
- Email support · 1 business day
- Credits valid 18 months
= 2.0¢ / image · 9.8¢ / video
24,500 credits · 1 credit per image, 5 per video
51% offProduction workloads, growing customer base.
- Everything in Builder
- 120 / min rate limit (2× baseline)
- Priority email support · <24h
- Credits valid 24 months
= 1.7¢ / image · 8.5¢ / video
100,000 credits · 1 credit per image, 5 per video
58% offEstablished platform, predictable monthly burn.
- Everything in Scale
- 240 / min rate limit (4× baseline)
- 5,000-char prompts (vs 2,000)
- Credits never expire
- Best self-serve rate
Custom commit, dedicated GPU pool (no shared throttling), bespoke moderation policy, multi-region failover, dedicated account manager. Volume-discount continues past the public tiers.
What each generation costs
Same credit-rate across all tiers; only the per-credit price changes by pack size.
Doing it yourself vs. plugging us in
A normal production setup runs three separate endpoints (realistic, anime and video) each on its own always-on GPU, since the models can't share one worker. That's a three-worker minimum at ~$1.50/hour = $3,240/month (idle or not), and prod-grade reliability needs more than one worker per endpoint as volume grows, for burst headroom and redundancy. Below: each plan's volume vs the equivalent DIY GPU bill.
~33 imgs/day · 1k/mo $3,240 / mo
3 workers · realistic + anime + video, idle $25 / mo
2.5¢ / image $3,215 / mo
~170 imgs/day · 5k/mo $6,480 / mo
6 workers · 2 per endpoint $110 / mo
2.2¢ / image $6,370 / mo
~830 imgs/day · 25k/mo $9,720 / mo
9 workers · 3 per endpoint $490 / mo
2.0¢ / image $9,230 / mo
~3,330 imgs/day · 100k/mo $12,960 / mo
12 workers · 4 per endpoint, burst $1,700 / mo
1.7¢ / image $11,260 / mo
Raw GPU rental only: excluded from the DIY column: ~3 months of engineering work (identity persistence, moderation, storage, webhooks, payments), content-moderation API spend ($5–$500/mo depending on volume), payment-processor cuts (adult merchants pay 6–10% vs Xavira's bundled rate), and ongoing ops: model updates, payment re-onboarding, cold-start tuning. Even on raw GPU alone the gap widens with volume (prod-grade burst needs more always-on workers per endpoint) and the 3-month build + permanent ops burden sits on top. Enterprise pricing ≤ 1.4¢ / image available from 500k credits.