Image generation
Anoman Studio — a guarded, OpenAI-images-compatible endpoint plus a full studio, both metered like any other usage.
POST /v1/images/generations
Generate images with an OpenAI-images-compatible call to POST /v1/images/generations using your API key. The models below are on the allowlist; an unknown model returns 400.
| Model | Class | Price / image | Plans |
|---|---|---|---|
flux-schnell | Fast (default) | Rp 71 | All plans |
flux-1.1-pro | Quality | Rp 924 | Plus Pass, Pro, Enterprise |
recraft-v3 | Quality | Rp 924 | Plus Pass, Pro, Enterprise |
flux-1.1-pro-ultra | Premium | Rp 1.386 | Pro, Enterprise |
ideogram-v3 | Premium | Rp 1.386 | Pro, Enterprise |
Prices are the provider's price for the image plus our margin, rounded up to the whole Rupiah, and are drawn from your prepaid balance. Prices shown are as of November 2026. The current price for each model is on the models page and in the studio. Free plan: 5 Fast images a day, up to 50 a month, at no charge with a verified email, while the monthly free image budget lasts. Passes include a monthly image allowance; past it, images draw your balance.
curl
response
Pass "response_format": "b64_json" to receive base64 image data, which you should do because the upstream provider URLs are ephemeral; otherwise the response contains those short-lived URLs. n is limited per request, and a per-key cap of 100 images/day applies.
Prompt enhancement
Set enhance: true to run a cheap LLM art-director that rewrites your prompt for quality and strips 'AI slop'. The guardrail then re-runs on the enhanced prompt before it reaches the image model, so enhancement can never sneak unsafe content through.
Add an optional style preset to steer the look:
product-photoeditorialflat-illustration3d-rendercinematic
The response surfaces the rewritten prompt at _anoman.enhanced_prompt and a per-item revised_prompt so you can see exactly what was sent to the model.
What's enforced
- The prompt passes the same media guard as video before anything is spent: child-safety (including ages written as numbers or words), violent extremism, NSFW and content moderation, in English and Bahasa Indonesia. The guard runs again on an enhanced prompt. Blocked prompts are never sent to a model and are not charged.
- Personal data the guard can validate (emails, phone numbers, ID numbers) is masked before the prompt leaves for the enhancer or the image model.
- Images are generated by fal.ai, outside Indonesia. Use
b64_jsonto keep the result; provider URLs are short-lived. - A per-key cap of 100 images/day and a 45-second per-generation timeout apply. The Free plan also has the 5-a-day Fast limit above.
- Your plan decides which models you can use (table above); a model above your plan returns 403
image_model_not_entitled.
One wallet. Image spend comes out of the same prepaid balance as text completions, with per-user attribution.
A designer UI
Prefer a UI? The Design studio lives at studio.anoman.io and inside the dashboard at /dashboard/design — a prompt composer with style chips, a model picker filtered to your plan, 1–4 images, palettes and boards, the price per run in your account's currency, a balance pill, aspect ratio, and download.
Both surfaces call the same guarded endpoint as your own user, so the per-user daily cap and metering apply identically whether you use the API or the studio.
Open the studio
Compose prompts, pick a style, and generate — all guarded and metered through your account.