Multimodal API

Generate speech, images, and videos using GPU-powered models. All endpoints consume credits and return base64-encoded results.

Base URL: https://api.ainative.studio/api/v1/multimodal

Authentication

All endpoints require authentication via API key or Bearer token.

POST /api/v1/multimodal/tts
Authorization: Bearer YOUR_TOKEN
Content-Type: application/json

---

Text-to-Speech (TTS)

Generate high-quality speech from text using MiniMax TTS.

POST/api/v1/multimodal/tts🔒

Cost: 14 credits per generation | Rate limit: 10/min

curl -X POST https://api.ainative.studio/api/v1/multimodal/tts \
  -H "Authorization: Bearer $TOKEN" \
  -H "Content-Type: application/json" \
  -d '{
    "text": "Welcome to AINative Studio.",
    "voice": "Wise_Woman"
  }'

import requests

response = requests.post(
    "https://api.ainative.studio/api/v1/multimodal/tts",
    headers={"Authorization": f"Bearer {token}"},
    json={
        "text": "Welcome to AINative Studio.",
        "voice": "Wise_Woman"
    }
)

data = response.json()
# data["audio_base64"] — base64-encoded MP3
# data["credits_used"] — credits consumed

Parameters:

Parameter	Type	Required	Default	Description
text	string	Yes	—	Text to synthesize (1-5000 chars)
voice	string	No	Wise_Woman	Voice profile

Response:

Field	Type	Description
audio_base64	string	Base64-encoded MP3 audio
format	string	Audio format (mp3)
duration_ms	int	Audio duration in milliseconds
usage_id	string	Usage record ID for tracking
credits_used	int	Credits consumed
credits_remaining	int	Remaining credit balance

---

Image Generation

Generate images from text prompts using Qwen Image Edit with optional LoRA styles.

POST/api/v1/multimodal/image🔒

Cost: 50 credits per image | Rate limit: 5/min

curl -X POST https://api.ainative.studio/api/v1/multimodal/image \
  -H "Authorization: Bearer $TOKEN" \
  -H "Content-Type: application/json" \
  -d '{
    "prompt": "A futuristic cityscape at sunset with neon lights",
    "width": 1024,
    "height": 1024
  }'

import requests, base64

response = requests.post(
    "https://api.ainative.studio/api/v1/multimodal/image",
    headers={"Authorization": f"Bearer {token}"},
    json={
        "prompt": "A futuristic cityscape at sunset with neon lights",
        "width": 1024,
        "height": 1024
    }
)

data = response.json()
# Save image
with open("output.png", "wb") as f:
    f.write(base64.b64decode(data["image_base64"]))

Parameters:

Parameter	Type	Required	Default	Description
prompt	string	Yes	—	Image description (1-2000 chars)
width	int	No	1024	Image width (512-2048)
height	int	No	1024	Image height (512-2048)
style	string	No	null	Optional LoRA style preset

Response:

Field	Type	Description
image_base64	string	Base64-encoded image
format	string	Image format (jpg or png)
width / height	int	Actual dimensions
credits_used	int	Credits consumed

---

Video: Image-to-Video (I2V)

Animate a static image into a video with motion prompts.

POST/api/v1/multimodal/video/i2v🔒

Rate limit: 2/min

Provider	Cost	Duration	Resolution	Quality
wan22 (default)	400 credits	5s	1280x720	Best value
seedance	520 credits	5s	1280x720	High quality
sora2 (premium)	800 credits	4s	1280x720	Cinematic

curl -X POST https://api.ainative.studio/api/v1/multimodal/video/i2v \
  -H "Authorization: Bearer $TOKEN" \
  -H "Content-Type: application/json" \
  -d '{
    "image_url": "https://example.com/photo.jpg",
    "motion_prompt": "Camera slowly zooms in while clouds drift across the sky",
    "provider": "wan22"
  }'

Parameters:

Parameter	Type	Required	Default	Description
image_url	string	Yes	—	Public URL of source image
motion_prompt	string	Yes	—	Motion description (1-1000 chars)
provider	string	No	wan22	wan22, seedance, or sora2

---

Video: Text-to-Video (T2V)

Generate a full video from a text description.

POST/api/v1/multimodal/video/t2v🔒

Cost: 1000 credits | Rate limit: 1/min | Tier: Premium only

curl -X POST https://api.ainative.studio/api/v1/multimodal/video/t2v \
  -H "Authorization: Bearer $TOKEN" \
  -H "Content-Type: application/json" \
  -d '{
    "prompt": "A serene ocean wave crashing on a beach at golden hour",
    "duration": 5
  }'

Parameters:

Parameter	Type	Required	Default	Description
prompt	string	Yes	—	Video description (1-1000 chars)
duration	int	No	5	Duration in seconds (1-10)

---

Video: CogVideoX

Generate video using the dedicated CogVideoX-2B model with fine-grained control.

POST/api/v1/multimodal/video/cogvideox🔒

Cost: 800 credits | Rate limit: 1/min

curl -X POST https://api.ainative.studio/api/v1/multimodal/video/cogvideox \
  -H "Authorization: Bearer $TOKEN" \
  -H "Content-Type: application/json" \
  -d '{
    "prompt": "A robot walking through a garden in the rain",
    "num_frames": 49,
    "guidance_scale": 6.0
  }'

Parameters:

Parameter	Type	Required	Default	Description
prompt	string	Yes	—	Video description (1-1000 chars)
num_frames	int	No	49	Frame count: 17, 33, or 49
guidance_scale	float	No	6.0	CFG scale (1.0-20.0)
num_inference_steps	int	No	50	Denoising steps (20-100)

Video Response (all video endpoints):

Field	Type	Description
video_url	string	URL to download generated video
format	string	Video format (mp4)
duration_seconds	float	Video duration
usage_id	string	Usage tracking ID
credits_used	int	Credits consumed
credits_remaining	int	Remaining balance

---

Usage History

Track your multimodal API usage across all endpoints.

GET/api/v1/multimodal/usage🔒

curl https://api.ainative.studio/api/v1/multimodal/usage?limit=10 \
  -H "Authorization: Bearer $TOKEN"

Query Parameters:

Parameter	Type	Default	Description
skip	int	0	Pagination offset
limit	int	50	Max records (1-100)
endpoint_type	string	—	Filter: tts, image, video_i2v, video_t2v, video_cogvideox

---

Credit Costs Summary

Endpoint	Credits	Approx. Cost
TTS	14	$0.007
Image	50	$0.025
I2V (wan22)	400	$0.20
I2V (seedance)	520	$0.26
I2V (sora2)	800	$0.40
T2V	1000	$0.50
CogVideoX	800	$0.40

Error Codes

Code	Meaning
402	Insufficient credits
400	Invalid request (bad params, unsupported format)
429	Rate limit exceeded
500	GPU provider error — retry after 30s

For AI Agents

Agents can use these endpoints to generate media autonomously. Best practices:

• Check credits first via GET /api/v1/managed/usage before starting expensive operations
• Use wan22 for I2V — best quality-to-cost ratio
• Poll usage records via GET /api/v1/multimodal/usage/{usage_id} to check generation status
• Base64 responses can be piped directly to file storage via POST /api/v1/public/zerodb/files/upload