If you’ve been digging through OpenAI’s API docs trying to figure out what GPT Image 2 will actually cost you, take a breath. You don’t need to memorize token tables, decode the “organization tier” rate limits, or set up a billing account. On imagesv2 you buy credits, and one image costs a known amount of credits before you click generate. This post is the boring-but-honest version of our pricing. No marketing math, no “unlimited” asterisks. Just what you pay, what you get, and how to pick the right plan.
Stop reading prices — generate one for free first
Sign up and get 3 free credits — enough for one image. No credit card.
[Try it free](/#playground)
[TL;DR — what you actually pay](#tldr--what-you-actually-pay)
| Plan | Yearly (best deal) | Monthly | Credits / month | Image budget* |
|---|---|---|---|---|
Starter | $4.16/mo ($49.9/year) | $8.40 | 1,000 (12,000/yr) | ~500 low-quality or ~120 high-quality |
Pro ⭐ | $16.66/mo ($199.9/year) | $33.90 | 5,000 (60,000/yr) | ~2,500 LQ or ~600 HQ | Ultra | $41.66/mo ($499.9/year) | $84.90 | 20,000 (240,000/yr) | ~10,000 LQ or ~2,400 HQ |
Rough estimates. High-quality 1024×1536 / 1536×1024 outputs cost more credits than 1024×1024 low-quality drafts. The exact credit cost is shown in the generator before you confirm.
Prefer not to subscribe? Buy credits once — no recurring charge, and credits never expire:
| One-time pack | Price | Credits | Per credit |
|---|---|---|---|
| Starter Pack | $9.90 | 500 | $0.0198 |
| Pro Pack | $39.90 | 5,000 | $0.008 |
Compared to subscriptions, one-time packs are slightly more per credit — that’s the trade-off for zero commitment. Most people who generate more than a few images per week save more on the yearly Pro plan.
Best to pick a plan after you've seen what it makes
Spend your 3 free credits on one image, then decide which plan fits.
[Generate one first](/#playground)
[How credits actually work](#how-credits-actually-work)
A credit is a small unit of generation budget. When you hit Generate, imagesv2 deducts credits based on three things:
Quality— Standard or High Quality. HQ uses more credits, but you can usually skip a regeneration cycle.** Size**— 1024×1024 (square) is cheapest. Wide and tall ratios (1024×1536, 1536×1024) cost a bit more because there’s more pixel area for the model to plan.Mode— generation costs more than editing an existing image. If you already have a good base, edit instead of regenerating from scratch.
The credit cost shows up in the studio before you confirm, so you never get surprised. No “you used your tokens” email three days later.
Pricing the way most people actually use it Almost every user falls into one of three patterns. Here’s how to think about it:
[“I just want to try it out” → Starter, monthly or one-time](#i-just-want-to-try-it-out--starter-monthly-or-one-time)
If you’re not sure whether GPT Image 2 fits your workflow yet, grab the **$9.90 one-time pack** (500 credits, no commitment) or the **Starter Monthly at $8.40**. You’ll get a real sense of the model — 60+ high-quality images is plenty to validate whether the look, the text rendering, and the editing flow match what you need. Cancel anytime; one-time credits don’t expire.
[“I’m using this for client work / daily creation” → Pro Yearly](#im-using-this-for-client-work--daily-creation--pro-yearly)
The Pro Yearly plan at $16.66/mo equivalent is where the math gets genuinely good: $199.9 buys 60,000 credits — roughly 2,400 high-quality images per year, or six per day. For freelance designers, indie founders, and content teams, this is the sweet spot. You also get the commercial license, priority queue, and watermark-free downloads.
“I’m running a content pipeline / agency” → Ultra Yearly Ultra Yearly ($499.9/year) lands at about $0.0021 per credit — half the per-credit cost of Starter. If you’re shipping hundreds of assets a month for clients, this is where it pays off. You also get the highest-priority queue (matters when everyone in your team is generating at the same time) and early access to new models.
Why credit-based pricing instead of token math? We thought hard about this when we built imagesv2 and chose credits for a specific reason: token-based pricing is unpredictable for non-engineers. With raw OpenAI billing, the cost depends on input token count, output token count, image dimensions, and quality multipliers — and you only see the total after the generation. Many users open their bill and find a $40 charge they can’t reverse-engineer.
Credits flip that around:
- You see the cost before you click generate.
- You buy in fixed packs, so spend is bounded.
- Yearly subscribers get the per-credit price locked in — no surprise repricing.
It’s a less “API-native” model, but it’s the one almost every creator we talk to prefers.
What you don’t pay for A few costs that other tools sneak in, and we don’t:
Watermark removal— included on every paid plan.** Commercial use**— included on Pro and Ultra. Use generated images in client work, products, and ads.** Extra fees per resolution**— you pay credits, full stop. No “HD upgrade” add-on at checkout.** API setup, org verification, or billing account hassle**— none of that. Sign in with Google and start generating.
A note on the “GPT Image 2 API tier limits” you might have read elsewhere Other articles online describe GPT Image 2’s rate limits in terms of OpenAI usage tiers (Tier 1 = 5 images/minute, etc.). That’s about the raw API — and if you’re building your own product on top of OpenAI’s API, those numbers matter to you.
If you’re using imagesv2 to *generate images*, they don’t. We pool capacity across plans and route requests through prioritized queues, so what you actually experience is a queue time per plan (Ultra → Pro → Starter), not a hard per-minute cap.
[Ready to try it?](#ready-to-try-it)
Generate first, pay later
3 free credits on signup. Watermark-free preview. Pick a plan when you're sold.
Make your first image Questions about whether your workflow fits a specific plan? Email us at [email protected] — we read every message.