Credit Costs
How many credits each model and variant costs per generation.
Every generation deducts credits from your team's balance at submission time. If a job fails, the credits are automatically refunded.
You can check your current balance via the Account endpoint or in the Apiframe console.
Image generation
| Model | Variant | Credits per generation |
|---|---|---|
| Midjourney | — | 5 |
| Nano Banana | — | 3 |
| Nano Banana Pro | 1K / 2K | 26 |
| Nano Banana Pro | 4K | 52 |
| Nano Banana 2 | 1K | 5 |
| Nano Banana 2 | 2K | 7 |
| Nano Banana 2 | 4K | 10 |
| Flux 1.1 Pro | — | 5 |
| Flux 1.1 Pro Ultra | — | 8 |
| Flux 2 Pro | — | 6 |
| Flux 2 Dev | — | 3 |
| Flux 2 Flex | — | 4 |
| Flux 2 Max | — | 10 |
| Seedream 3 | — | 3 |
| Seedream 3.1 | — | 3 |
| Seedream 4 | — | 3 |
| Seedream 4.5 | — | 4 |
| Seedream 5 Lite | — | 4 |
| Ideogram V2 | — | 8 |
| Ideogram V2 Turbo | — | 5 |
| Ideogram V2a | — | 4 |
| Ideogram V2a Turbo | — | 3 |
| Ideogram V3 Balanced | — | 6 |
| Ideogram V3 Turbo | — | 3 |
| Ideogram V3 Quality | — | 9 |
| Ideogram Character | Turbo | 10 |
| Ideogram Character | Default | 15 |
| Ideogram Character | Quality | 20 |
| Imagen 3 | — | 5 |
| Imagen 3 Fast | — | 3 |
| Imagen 4 | — | 4 |
| Imagen 4 Ultra | — | 6 |
| Imagen 4 Fast | — | 2 |
| DALL-E 3 | — | 12 |
| DALL-E 2 | — | 2 |
| GPT Image 1.5 | Auto | 14 × n |
| GPT Image 1.5 | Low | 2 × n |
| GPT Image 1.5 | Medium | 5 × n |
| GPT Image 1.5 | High | 14 × n |
| GPT Image 2 | Low | 4 × n |
| GPT Image 2 | Medium | 9 × n |
| GPT Image 2 | High | 23 × n |
| GPT Image 2 | Auto | 23 × n |
| Kling Image | — | 2 × n |
| Qwen Image | — | 3 × n |
| Qwen Image 2 | — | 3 |
| Qwen Image 2 Pro | — | 10 |
| Grok Imagine Image | — | 3 |
| Reve Create | — | 3 |
| Flux LoRA (AI Photos inference) | — | 2 × num_outputs |
AI Photos (LoRA training)
POST /v2/loras — train a Flux LoRA from 15-30 subject photos.
| Step | Credits | Notes |
|---|---|---|
| LoRA create (upload + caption + zip) | 2 | Charged at submission. Refunded on failure. |
| LoRA training | 255 | Charged just before training starts. Refunded if training fails. |
| Inference per image | 2 | See Flux LoRA — num_outputs of 1-4 supported. |
Video generation
| Model | Variant | Credits per generation |
|---|---|---|
| Kling 2.6 Pro | 5s (no audio) | 15 |
| Kling 2.6 Pro | 5s (with audio) | 25 |
| Kling 2.6 Pro | 10s (no audio) | 25 |
| Kling 2.6 Pro | 10s (with audio) | 40 |
Wan 2.7 (per second of generated video)
credits = rate × duration (default duration = 5 s).
| Model | Variant | Credits / second |
|---|---|---|
| Wan 2.7 | T2V · 720p | 17 |
| Wan 2.7 | T2V · 1080p | 17 |
| Wan 2.7 | I2V · 720p | 17 |
| Wan 2.7 | I2V · 1080p | 26 |
| Wan 2.7 R2V | any resolution | 17 |
Image editing
POST /v2/images/upscale, POST /v2/images/background-remove, POST /v2/images/edit — Upscale, background removal, inpainting and outpainting.
| Model | Variant | Credits per generation |
|---|---|---|
| Topaz Image Upscale | ≤ 12 MP output | 9 |
| Topaz Image Upscale | ≤ 24 MP output | 9 |
| Topaz Image Upscale | ≤ 36 MP output | 17 |
| Topaz Image Upscale | ≤ 48 MP output | 17 |
| Topaz Image Upscale | ≤ 60 MP output | 26 |
| Topaz Image Upscale | ≤ 96 MP output | 34 |
| Topaz Image Upscale | ≤ 132 MP output | 41 |
| Topaz Image Upscale | ≤ 168 MP output | 50 |
| Topaz Image Upscale | ≤ 336 MP output | 91 |
| Topaz Image Upscale | ≤ 512 MP output | 140 |
| Clarity Upscaler | — | 3 |
| Bria Background Remove | — | 4 |
| 851-labs Background Remove | — | 2 |
| Flux Fill Pro (inpaint or outpaint) | — | 9 |
The Topaz output-megapixel tier is computed automatically from the input image's dimensions × upscale_factor² and rounded up to the next configured row.
Video editing
POST /v2/videos/upscale — Topaz video upscale and frame interpolation.
Pricing is per second of input video, tiered by output resolution × output fps. The route probes the input duration via ffprobe at submit time and charges cost-per-second × ceil(duration).
| Model | Variant | Credits / second of input |
|---|---|---|
| Topaz Video Upscale | 720p · 24 fps | 1 |
| Topaz Video Upscale | 720p · 30 fps | 1 |
| Topaz Video Upscale | 720p · 60 fps | 2 |
| Topaz Video Upscale | 1080p · 24 fps | 4 |
| Topaz Video Upscale | 1080p · 30 fps | 4 |
| Topaz Video Upscale | 1080p · 60 fps | 7 |
| Topaz Video Upscale | 4K · 24 fps | 13 |
| Topaz Video Upscale | 4K · 30 fps | 13 |
| Topaz Video Upscale | 4K · 60 fps | 26 |
| Wan 2.7 VideoEdit | any resolution | 17 |
Music generation
| Model | Credits per generation |
|---|---|
| Suno | 10 |
How credits work
- Deducted at submission — credits are reserved the moment you send a generation request.
- Refunded on failure — if a job fails (provider error, content policy violation, etc.), the full credit cost is returned to your balance.
- Snapshotted per job — the
creditCostfield on a job reflects the cost at the time of submission, even if pricing changes later. - Team-scoped — credits belong to your team, not individual users. All team members share the same balance.
Checking your balance
curl https://api.apiframe.ai/v2/me \
-H "X-API-Key: afk_your_api_key_here"The response includes your current credit balance:
{
"id": "...",
"email": "...",
"team": {
"id": "...",
"credits": 850
}
}