HunyuanImage-3.0
HunyuanImage-3.0 is an 80B parameter MoE model for high fidelity text to image generation. It uses an autoregressive multimodal framework for strong world knowledge reasoning and sharp text rendering. It targets complex long prompts and precise layout control for production workloads.
API Options
Platform-level options for task execution and delivery.
-
taskType
string required value: imageInference -
Identifier for the type of task being performed
-
taskUUID
string required UUID v4 -
UUID v4 identifier for tracking tasks and matching async responses. Must be unique per task.
-
outputType
string default: URL -
Image output type.
Allowed values 3 values
-
outputFormat
string default: JPG -
Specifies the file format of the generated output. The available values depend on the task type and the specific model's capabilities.
- `JPG`: Best for photorealistic images with smaller file sizes (no transparency).
- `PNG`: Lossless compression, supports high quality and transparency (alpha channel).
- `WEBP`: Modern format providing superior compression and transparency support.
**Transparency**: If you are using features like background removal or LayerDiffuse that require transparency, you must select a format that supports an alpha channel (e.g., `PNG`, `WEBP`, `TIFF`). `JPG` does not support transparency.Allowed values 3 values
-
outputQuality
integer min: 20 max: 99 default: 95 -
Compression quality of the output. Higher values preserve quality but increase file size.
-
webhookURL
string URI -
Specifies a webhook URL where JSON responses will be sent via HTTP POST when generation tasks complete. For batch requests with multiple results, each completed item triggers a separate webhook call as it becomes available.
Learn more 1 resource
- Webhooks PLATFORM
- Webhooks
-
deliveryMethod
string default: sync -
Determines how the API delivers task results.
Allowed values 2 values
- Returns complete results directly in the API response.
- Returns an immediate acknowledgment with the task UUID. Poll for results using getResponse.
Learn more 1 resource
- Task Polling PLATFORM
-
uploadEndpoint
string URI -
Specifies a URL where the generated content will be automatically uploaded using the HTTP PUT method. The raw binary data of the media file is sent directly as the request body. For secure uploads to cloud storage, use presigned URLs that include temporary authentication credentials.
Common use cases:
- Cloud storage: Upload directly to S3 buckets, Google Cloud Storage, or Azure Blob Storage using presigned URLs.
- CDN integration: Upload to content delivery networks for immediate distribution.
// S3 presigned URL for secure upload https://your-bucket.s3.amazonaws.com/generated/content.mp4?X-Amz-Signature=abc123&X-Amz-Expires=3600 // Google Cloud Storage presigned URL https://storage.googleapis.com/your-bucket/content.jpg?X-Goog-Signature=xyz789 // Custom storage endpoint https://storage.example.com/uploads/generated-image.jpgThe content data will be sent as the request body to the specified URL when generation is complete.
-
safety
object -
Content safety checking configuration for image generation.
Properties 2 properties
-
safety»checkContentcheckContent
boolean default: false -
Enable or disable content safety checking. When enabled, defaults to
fastmode.
-
safety»modemode
string default: none -
Safety checking mode for image generation.
Allowed values 2 values
- Disables checking.
- Performs a single check.
-
-
ttl
integer min: 60 -
Time-to-live (TTL) in seconds for generated content. Only applies when
outputTypeisURL.
-
includeCost
boolean default: false -
Include task cost in the response.
-
numberResults
integer min: 1 max: 20 default: 1 -
Number of results to generate. Each result uses a different seed, producing variations of the same parameters.
-
acceleration
string default: medium -
Optimization level.
Allowed values 2 values
-
acceleratorOptions
object -
Advanced caching mechanisms to speed up generation.
Properties 9 properties
-
acceleratorOptions»cacheEndStepcacheEndStep
integer min: 1 -
Absolute step number to end caching. Must be greater than
cacheStartStepand less than or equal tosteps.
-
acceleratorOptions»cacheEndStepPercentagecacheEndStepPercentage
integer min: 1 max: 100 -
Percentage of steps to end caching. Alternative to
cacheEndStep. Must be greater thancacheStartStepPercentage.
-
acceleratorOptions»cacheMaxConsecutiveStepscacheMaxConsecutiveSteps
integer min: 1 max: 5 default: 3 -
Limits the maximum number of consecutive steps that can use cached computations before forcing a fresh computation.
-
acceleratorOptions»cacheStartStepcacheStartStep
integer min: 0 -
Absolute step number to start caching. Must be less than
cacheEndStep.
-
acceleratorOptions»cacheStartStepPercentagecacheStartStepPercentage
integer min: 0 max: 99 -
Percentage of steps to start caching. Alternative to
cacheStartStep. Must be less thancacheEndStepPercentage.
-
acceleratorOptions»fbCachefbCache
boolean default: false -
First Block Cache (FBCache) acceleration. Reuses feature block computations across steps.
-
acceleratorOptions»fbCacheThresholdfbCacheThreshold
float min: 0 max: 1 step: 0.01 default: 0.25 -
Controls the sensitivity threshold for determining when to reuse cached computations. Lower values reuse more aggressively.
-
acceleratorOptions»teaCacheteaCache
boolean default: false -
TeaCache acceleration for transformer-based models. Estimates step differences to skip redundant computations.
-
acceleratorOptions»teaCacheDistanceteaCacheDistance
float min: 0 max: 1 step: 0.01 default: 0.5 -
Controls the aggressiveness of the TeaCache feature. Lower values prioritize quality, higher values prioritize speed.
-
Generation Parameters
Core parameters for controlling the generated content.
-
model
string required value: runware:180@1 -
Identifier of the model to use for generation.
Learn more 3 resources
-
positivePrompt
string required min: 2 max: 3000 -
Text prompt describing elements to include in the generated output.
Learn more 2 resources
-
negativePrompt
string min: 2 max: 3000 -
Prompt to guide what to exclude from generation. Ignored when guidance is disabled (CFGScale ≤ 1).
Learn more 1 resource
-
Width of the generated media in pixels.
Learn more 2 resources
-
Height of the generated media in pixels.
Learn more 2 resources
-
seed
integer min: 0 max: 9223372036854776000 -
Random seed for reproducible generation. When not provided, a random seed is generated in the unsigned 32-bit range.
Learn more 1 resource
-
steps
integer min: 1 max: 100 default: 20 -
Total number of denoising steps. Higher values generally produce more detailed results but take longer.
Learn more 1 resource
-
scheduler
string -
Scheduler to use for the diffusion process.
Allowed values 75 values
Learn more 2 resources
-
CFGScale
float min: 1.1 max: 50 step: 0.01 default: 7 -
Guidance scale representing how closely the output will resemble the prompt. Higher values produce results more aligned with the prompt.
Learn more 1 resource
Submerged Cathedral Transit Hall
{
"taskType": "imageInference",
"taskUUID": "c3f77111-0f44-4286-823a-e5839cd877cf",
"model": "runware:180@1",
"positivePrompt": "A vast underwater transit concourse built inside a flooded stone cathedral, viewed from a low wide-angle perspective. Ribbed gothic arches and fractured stained glass rise above polished platforms, while clear seawater fills the upper half of the hall beyond transparent pressure walls. Sleek commuter trams wait on two parallel tracks, their design elegant and plausible, blending advanced engineering with salvaged historic architecture. Schools of silver fish drift through shafts of filtered sunlight, casting moving caustic patterns across marble floors and brass handrails. Travelers in practical weatherproof clothing walk between kiosks, luggage carts, and illuminated route boards. Include crisp readable station signage that says 'ABYSSAL TERMINAL' above the central platform and a smaller directional sign reading 'TIDE LINE 4'. The composition should feel grand yet believable, with layered depth from foreground benches and puddles to distant arches fading into blue water haze. Materials should be richly rendered: wet stone, oxidized metal, glass, ceramic tile, fabric, and bubbles trapped in seams. Mood: awe, calm, and quiet human resilience. Style: cinematic hyper-detailed environmental concept image with realistic lighting, sharp focus, nuanced textures, controlled color harmony of teal, gold, ivory, and deep marine blue.",
"negativePrompt": "low detail, blurry, distorted anatomy, extra limbs, duplicated people, warped tram geometry, unreadable text, misspelled signage, flat lighting, oversaturated colors, cartoon, low-resolution, cropped subjects, empty scene, chaotic composition",
"width": 1536,
"height": 512,
"seed": 96688,
"steps": 32,
"CFGScale": 7.5
}{
"taskType": "imageInference",
"taskUUID": "c3f77111-0f44-4286-823a-e5839cd877cf",
"imageUUID": "1dae5b0f-eb48-40bd-8cb5-0dc8c9d9f1a4",
"imageURL": "https://im.runware.ai/image/ws/2/ii/1dae5b0f-eb48-40bd-8cb5-0dc8c9d9f1a4.jpg",
"seed": 96688,
"cost": 0.1453
}Floating Archive Balloon Regatta
{
"taskType": "imageInference",
"taskUUID": "b5e41b0b-cb66-434d-90fb-58af2d15d175",
"model": "runware:180@1",
"positivePrompt": "A sweeping fantasy panorama of a sky regatta made of enormous scholarly hot-air balloons drifting above a patchwork valley at sunrise. Each balloon is uniquely designed as a flying archive: one wrapped in brass rings and parchment sails, one shaped like a giant glass inkwell with suspended reading decks, one built from lacquered wood with tiny windows and hanging rope bridges, one carrying rotating armillary-like fins and stacked map cases. In the foreground, a broad airborne docking platform of timber and riveted metal holds travelers, rolling trunks, crates of scroll tubes, fluttering pennants, and a large painted sign that clearly reads SKYWARD RECORD EXCHANGE. Midground: dozens of smaller balloons and gliders weave through the air lanes in layered depth. Background: terraced farms, winding rivers, distant villages, and soft mountain ridges fading into atmospheric haze. Golden low-angle light, voluminous clouds, highly detailed materials, cinematic scale, believable perspective, intricate worldbuilding, sharp focus, clean text rendering, vivid but natural colors.",
"negativePrompt": "blurry, low detail, distorted anatomy, cropped balloons, duplicate aircraft, illegible text, warped sign lettering, oversaturated, flat lighting, empty sky, modern skyscrapers, futuristic holograms, underwater scene",
"width": 1280,
"height": 768,
"seed": 8052,
"steps": 30,
"CFGScale": 7.5
}{
"taskType": "imageInference",
"taskUUID": "b5e41b0b-cb66-434d-90fb-58af2d15d175",
"imageUUID": "c956cd64-2650-4828-84c1-6a1f0b4da0e7",
"imageURL": "https://im.runware.ai/image/ws/2/ii/c956cd64-2650-4828-84c1-6a1f0b4da0e7.jpg",
"seed": 8052,
"cost": 0.1735
}Amber Tidal Foundry Procession
{
"taskType": "imageInference",
"taskUUID": "9faa0571-9d4f-408d-869e-e017de74583a",
"model": "runware:180@1",
"positivePrompt": "A vast tidal foundry built inside a colossal coastal drydock at low tide, viewed in a wide cinematic composition. In the foreground, a ceremonial procession of brass-and-ceramic tool bearers crosses shallow reflective water on temporary steel grates. In the midground, towering gantries suspend half-assembled cargo skiffs with exposed ribs, pulley lines, riveted panels, and glowing furnace mouths casting warm amber light through steam. In the background, terraced workshops are carved into the harbor wall, packed with ladders, cranes, coiled hoses, stacked molds, and workers in weathered waterproof aprons. A giant painted sign over the main forge reads clearly: \"FOUNDRY NINE\". Another smaller wayfinding placard near the grates reads: \"TIDE PATH B\". The sky is overcast after a storm has passed, with breaks of pale sunlight striking wet metal and rippled mudflats. Atmosphere of industrious ritual and monumental craftsmanship, richly textured surfaces, realistic materials, sharp focus, deep perspective, intricate details, cinematic lighting, subtle haze, natural color grading, highly coherent spatial layout.",
"negativePrompt": "low resolution, blurry, distorted anatomy, duplicated people, extra limbs, warped text, unreadable signage, oversaturated colors, empty scene, minimal detail, cartoon, flat lighting, fisheye distortion",
"width": 1472,
"height": 576,
"seed": 38034,
"steps": 30,
"CFGScale": 6.5
}{
"taskType": "imageInference",
"taskUUID": "9faa0571-9d4f-408d-869e-e017de74583a",
"imageUUID": "b9839c5d-cb14-4402-bfda-ef8e68d00f0b",
"imageURL": "https://im.runware.ai/image/ws/2/ii/b9839c5d-cb14-4402-bfda-ef8e68d00f0b.jpg",
"seed": 38034,
"cost": 0.1491
}Rain-Soaked Orbital Train Concourse
{
"taskType": "imageInference",
"taskUUID": "74bc041b-53e7-415d-bb98-3c433bc30efa",
"model": "runware:180@1",
"positivePrompt": "A vast orbital train concourse inside a rotating ring station, seen from a slightly elevated three-quarter view. Curved panoramic windows reveal a blue gas giant and thin bands of debris outside. The station floor is wet from tracked-in rain and melting frost, creating glossy reflections across dark stone tiles. Sleek high-speed trains with brushed titanium bodies wait at two platforms, doors open, soft cabin light spilling out. Diverse travelers move through the hall: a family with stacked luggage cubes, maintenance workers in weatherproof orange suits, a violinist under a glass canopy, two children pressing their hands to the window, and a courier pushing a cart of sealed parcels. Overhead hangs a large wayfinding board with crisp readable text: 'RING LINE A', 'POLAR EXPRESS', 'GATE 12', and a smaller illuminated sign reading 'ARRIVALS'. Tall structural ribs arc across the ceiling, threaded with cables, mist, and subtle condensation. Small kiosks sell maps and travel badges, with neatly arranged displays but no crowds blocking the architecture. Atmosphere of weary wonder, grounded science fiction, realistic materials, intricate reflections, precise perspective, sharp focus, richly detailed midground and background, dramatic volumetric light through rain streaks on the glass, cinematic color contrast of cool steel blues and warm amber highlights, ultra-detailed, believable scale, premium concept render quality.",
"negativePrompt": "low resolution, blurry, distorted anatomy, duplicate people, extra limbs, unreadable text, oversaturated colors, flat lighting, empty scene, floating objects, warped perspective, cropped trains, messy composition, heavy fog obscuring details, cartoon style, painterly brushstrokes",
"width": 1344,
"height": 704,
"seed": 19568,
"steps": 30,
"CFGScale": 7.5
}{
"taskType": "imageInference",
"taskUUID": "74bc041b-53e7-415d-bb98-3c433bc30efa",
"imageUUID": "a09ff7d8-c814-4fd9-b721-0c328908f70e",
"imageURL": "https://im.runware.ai/image/ws/2/ii/a09ff7d8-c814-4fd9-b721-0c328908f70e.jpg",
"seed": 19568,
"cost": 0.1671
}