Skip to content

CreoHub API

One API for text, video and images, avatars and speech

91
models in the catalog
6
generation categories
1
key and a shared balance
0
charged for a failed run

Catalog

All models and prices

Prices are per unit: a second of video, an image, a megapixel, or 1M tokens. The exact cost of any request is available before it starts via POST /v1/estimate.

Text10

OpenAI4
  • GPT-6 Astraopenai/gpt-6-astraup to 1 050 000 context tokens$30per 1M input$75per 1M output
  • GPT-6 Astra Proopenai/gpt-6-astra-proup to 1 050 000 context tokens$30per 1M input$75per 1M output
  • GPT-5.4openai/gpt-5.4up to 256 000 context tokens$3.75per 1M input$22.5per 1M output
  • GPT-4.1openai/gpt-4.1up to 1 047 576 context tokens$3per 1M input$12per 1M output
Anthropic2
  • Claude Sonnet 4.6anthropic/claude-sonnet-4.6up to 200 000 context tokens$4.5per 1M input$22.5per 1M output
  • Claude Opus 4.6anthropic/claude-opus-4.6up to 200 000 context tokens$7.5per 1M input$37.5per 1M output
Gemini2
  • Gemini 2.5 Progoogle/gemini-2.5-proup to 1 000 000 context tokens$1.88per 1M input$15per 1M output
  • Gemini 2.5 Flashgoogle/gemini-2.5-flashup to 1 000 000 context tokens$0.45per 1M input$3.75per 1M output
DeepSeek1
  • DeepSeek V3deepseek/deepseek-chat-v3-0324up to 128 000 context tokens$0.375per 1M input$1.5per 1M output
Llama1
  • Llama 4 Maverickmeta-llama/llama-4-maverickup to 128 000 context tokens$0.3per 1M input$1.2per 1M output

Video54

Kling18
  • Kling v3 Prokling-video/v3/pro/text-to-videoup to 15 s$0.21per second
  • Kling v3 Pro · imagekling-video/v3/pro/image-to-videoup to 15 s$0.21per second
  • Kling v3 Standardkling-video/v3/standard/text-to-videoup to 15 s$0.21per second
  • Kling v3 Standard · imagekling-video/v3/standard/image-to-videoup to 15 s$0.21per second
  • Kling v3 4Kkling-video/v3/4k/text-to-videoup to 15 s$0.63per second
  • Kling v3 4K · imagekling-video/v3/4k/image-to-videoup to 15 s$0.63per second
  • Kling 2.6 Prokling-video/v2.6/pro/text-to-videoup to 10 s$0.105per second
  • Kling 2.6 Pro · imagekling-video/v2.6/pro/image-to-videoup to 10 s$0.105per second
  • Kling 2.5 Turbo Prokling-video/v2.5-turbo/pro/text-to-videoup to 10 s$0.105per second
  • Kling 2.5 Turbo Pro · imagekling-video/v2.5-turbo/pro/image-to-videoup to 10 s$0.105per second
  • Kling O3 Prokling-video/o3/pro/text-to-videoup to 15 s$0.21per second
  • Kling O3 Pro · imagekling-video/o3/pro/image-to-videoup to 15 s$0.21per second
  • Kling v3 Pro Motion Controlkling-video/v3/pro/motion-controlup to 30 s$0.252per second
  • Kling 2.6 Motion Controlkling-video/v2.6/standard/motion-controlup to 30 s$0.105per second
  • Kling O1 Edit Standardkling-video/o1/standard/video-to-video/editup to 30 s$0.189per second
  • Kling O1 Edit Prokling-video/o1/video-to-video/editup to 30 s$0.252per second
  • Kling O3 Pro Editkling-video/o3/pro/video-to-video/editup to 30 s$0.21per second
  • Kling O3 4K Editkling-video/o3/4k/video-to-video/editup to 30 s$0.63per second
Google Veo8
  • Veo 3.1veo3.1up to 8 s$0.6per second
  • Veo 3.1 · imageveo3.1/image-to-videoup to 8 s$0.6per second
  • Veo 3.1 Fastveo3.1/fastup to 8 s$0.225per second
  • Veo 3.1 Fast · imageveo3.1/fast/image-to-videoup to 8 s$0.225per second
  • Veo 3.1 Liteveo3.1/liteup to 8 s$0.075per second
  • Veo 3.1 Lite · imageveo3.1/lite/image-to-videoup to 8 s$0.075per second
  • Veo 3.1 Referenceveo3.1/reference-to-videoup to 8 s$0.6per second
  • Veo 3.1 Fast Referenceveo3.1/fast/reference-to-videoup to 8 s$0.225per second
Seedance8
  • Seedance 2.0bytedance/seedance-2.0/text-to-videoup to 15 s480p720p1080p4k$1.02per second
  • Seedance 2.0 · imagebytedance/seedance-2.0/image-to-videoup to 15 s480p720p1080p4k$1.02per second
  • Seedance 2.0 Fastbytedance/seedance-2.0/fast/text-to-videoup to 15 s480p720p1080p$0.363per second
  • Seedance 2.0 Fast · imagebytedance/seedance-2.0/fast/image-to-videoup to 15 s480p720p1080p$0.363per second
  • Seedance 2.0 Referencebytedance/seedance-2.0/reference-to-videoup to 15 s480p720p1080p4k$1.02per second
  • Seedance 2.5bytedance/seedance-2.5/text-to-videoup to 30 s480p720p1080p$0.693per second
  • Seedance 2.5 · imagebytedance/seedance-2.5/image-to-videoup to 30 s480p720p1080p$0.693per second
  • Seedance 2.5 Referencebytedance/seedance-2.5/reference-to-videoup to 30 s480p720p1080p$0.693per second
Grok Imagine5
  • Grok Imagine Videoxai/grok-imagine-video/text-to-videoup to 15 s480p720p$0.105per second
  • Grok Imagine Video · imagexai/grok-imagine-video/image-to-videoup to 15 s480p720p$0.105per second
  • Grok Imagine Video 1.5xai/grok-imagine-video/v1.5/text-to-videoup to 15 s480p720p1080p$0.21per second
  • Grok Imagine Video 1.5 · imagexai/grok-imagine-video/v1.5/image-to-videoup to 15 s480p720p1080p$0.225per second
  • Grok Imagine Video 1.5 · referencexai/grok-imagine-video/v1.5/reference-to-videoup to 15 s480p720p1080p$0.21per second
Happy Horse4
  • Happy Horse 1.0alibaba/happy-horse/text-to-videoup to 15 s720p1080p$0.42per second
  • Happy Horse 1.0 · imagealibaba/happy-horse/image-to-videoup to 15 s720p1080p$0.42per second
  • Happy Horse 1.1alibaba/happy-horse/v1.1/text-to-videoup to 15 s720p1080p$0.27per second
  • Happy Horse 1.1 · imagealibaba/happy-horse/v1.1/image-to-videoup to 15 s720p1080p$0.27per second
Sora4
  • Sora 2sora-2/text-to-videoup to 20 s$0.15per second
  • Sora 2 · imagesora-2/image-to-videoup to 20 s$0.15per second
  • Sora 2 Prosora-2/text-to-video/proup to 20 s720p1080p$0.45per second
  • Sora 2 Pro · imagesora-2/image-to-video/proup to 20 s720p1080p$0.45per second
Wan4
  • Wan 2.7wan/v2.7/text-to-videoup to 15 s$0.15per second
  • Wan 2.7 · imagewan/v2.7/image-to-videoup to 15 s$0.15per second
  • Wan 3.0 Primealibaba/wan-3.0-prime/text-to-videoup to 30 s480p720p1080p$0.21per second
  • Wan 3.0 Prime · imagealibaba/wan-3.0-prime/image-to-videoup to 30 s480p720p1080p$0.21per second
FLUX2
  • FLUX 3 Videoblackforestlabs/flux-3/text-to-videoup to 20 s720p1080p$0.255per second
  • FLUX 3 Video · imageblackforestlabs/flux-3/image-to-videoup to 20 s720p1080p$0.255per second
Luma1
  • Luma Ray 2 Reframeluma-dream-machine/ray-2/reframeup to 30 s$0.3per second

Images14

Seedream3
  • Seedream 5 Litebytedance/seedream/v5/lite/editup to 4 pcs$0.052per image
  • Seedream 5 Lite · textbytedance/seedream/v5/lite/text-to-imageup to 4 pcs$0.052per image
  • Seedream 5 Probytedance/seedream/v5/pro/text-to-imageup to 4 pcs480p720p1080p4k$0.203per image
Nano Banana2
  • Nano Banana Pronano-banana-proup to 4 pcs$0.225per image
  • Nano Banana 2nano-banana-2up to 4 pcs0.5k1k2k4k$0.12per image
Qwen2
  • Qwen Imageqwen-imageup to 16 MP$0.03per megapixel
  • Qwen Image Editqwen-image-editup to 16 MP$0.045per megapixel
Recraft2
  • Recraft V4 Prorecraft/v4/pro/text-to-imageup to 4 pcs$0.375per image
  • Recraft V4 Vectorrecraft/v4/text-to-vectorup to 4 pcs$0.12per image
BiRefNet1
  • BiRefNetbirefnetup to 1 pcs$0.03per image
FLUX1
  • FLUX 2 Proflux-2-proup to 16 MP$0.045per megapixel
GPT Image1
  • GPT Image 2openai/gpt-image-2up to 4 pcs$0.08per image
Grok Imagine1
  • Grok Imagine Imagexai/grok-imagine-image/quality/text-to-imageup to 4 pcs$0.075per image
HiDream1
  • HiDream O1hidream-o1-imageup to 16 MP$0.015per megapixel

Avatars and lip-sync6

Sync Lipsync3
  • Sync Lipsync 2sync-lipsync/v2up to 60 s$0.075per second
  • Sync Lipsync 2 Prosync-lipsync/v2/proup to 60 s$0.125per second
  • Sync Lipsync 3sync-lipsync/v3up to 60 s$0.2per second
Kling Avatar1
  • Kling Avatar v2kling-video/ai-avatar/v2/standardup to 30 s$0.084per second
LatentSync1
  • LatentSynclatentsyncup to 60 s$0.0075per second
OmniHuman1
  • OmniHuman 1.5bytedance/omnihuman/v1.5up to 30 s$0.084per second

Upscale4

Topaz4
  • Topaz Image Upscaletopaz/upscale/imageup to 4 pcs$0.12per image
  • Topaz Image Generativetopaz/upscale/image/generativeup to 4 pcs$0.72per image
  • Topaz Video Upscaletopaz/upscale/videoup to 60 s$0.12per second
  • Topaz Video Generativetopaz/upscale/video/generativeup to 60 s$0.39per second

Speech and audio3

Demucs1
  • Demucsdemucsup to 3600 s$0.001per second
ElevenLabs1
  • Scribe v2elevenlabs/speech-to-text/scribe-v2up to 3600 s$0.0002per second
Whisper1
  • Whisperwhisperup to 3600 s$0.0019per second

Integration

Three steps to the first result

  1. 1

    Key

    The Default key appears right after sign-up. It runs on the same balance as the CreoHub studio — no separate top-up needed.

  2. 2

    Request

    POST /v1/queue/{model_id} with the model input. The response holds request_id and links to the status, the result and cancellation.

  3. 3

    Result

    Poll status_url or pass webhook_url. Finished files are served through signed CreoHub links valid for 24 hours.

curl
curl -X POST https://api.creohub.ai/v1/queue/veo3.1 \
  -H "Authorization: Key ch_live_…" \
  -H "Content-Type: application/json" \
  -d '{"prompt":"a quiet street at dusk, handheld","duration":"8","resolution":"1080p","generate_audio":true}'

Billing

Transparent charges, no surprises

Price before the run

POST /v1/estimate returns the exact cost of a request. The catalog lists prices per second, image, megapixel or 1M tokens.

Reserve, then the actual charge

The amount is reserved when the job is queued. After the result the actual cost is charged — never above the estimate.

Failure — no charge

If a generation fails, the reserve returns to the balance automatically. You pay only for results.

Spend per key

Separate keys for projects and teams, a monthly limit per key and a per-model breakdown in the console.

For AI agents

Ready for an agent to plug in

A machine-readable guide with the live catalog, an OpenAPI schema and a JSON price list. One link is enough for an agent.

Account

Access keys

After sign-up the Default key is already waiting in the console. Catalog, prices and docs are open without an account.

Reliability

What a team gets

  • One queue format for every model: status, result, cancellation
  • A completion webhook or status polling — your choice
  • Spend by key and model, monthly limits, request log
  • Result files are not archived — only signed links valid for 24 hours

FAQ

Frequently asked questions

How is the cost of a request calculated?

By the model unit: a second of video, an image, a megapixel, or 1M tokens. Without duration the estimate uses the model maximum; a missing num_images counts as 1. For models priced by resolution, pass resolution. For text, pass max_tokens.

Does the API need a separate top-up?

No. The API and the CreoHub studio share one account balance. Every operation shows in the wallet tagged API.

How long are results kept?

A signed link is valid for 24 hours and then returns 410. Download files as soon as the result arrives.

How do I upload an input file?

POST /v1/files with the bytes and a Content-Type. The response holds a file_id like chfile_…, which goes into any URL field of the input.

How do I connect the API to my own agent?

Give the agent https://api.creohub.ai/v1/llms.txt. It has the live catalog and the queue and text request format. The catalog and prices at https://creohub.ai/en/developers are open without an account.