Civitai Developer

Appearance

Chroma / ERNIE / Qwen / Z-Image LoRA training ​

Five smaller image-LoRA ecosystems share this page: each has its own ecosystem value and base checkpoint, but the request shape is otherwise the AI Toolkit standard.

ecosystemBaseDefault priceBest for
chromalodestones/Chroma1-HD2000 BuzzChroma community model fine-tunes
erniebaidu/ERNIE-Image1000 BuzzERNIE Image LoRAs
qwenQwen-Image (versioned)2000 BuzzQwen Image / Qwen-Image-Edit LoRAs
zimageturboostris/Z-Image-De-Turbo (+ Z-Image-Turbo extras)1000 BuzzZ-Image Turbo LoRAs (cheap, fast inference)
zimagebaseTongyi-MAI/Z-Image1000 BuzzZ-Image base LoRAs

Each ecosystem has its own subsection with a runnable example. The shared schema lives in Common parameters; ecosystem-specific quirks are in each subsection.

Long-running step

Always submit with wait=0. These ecosystems run anywhere from a fraction of a second per step (Z-Image Turbo) to ~1s/step (Chroma/Qwen). See Results & webhooks.

The request shape ​

json
{
  "$type": "training",
  "input": {
    "engine":    "ai-toolkit",
    "ecosystem": "chroma"   // chroma | ernie | qwen | zimageturbo | zimagebase
  }
}

Prerequisites ​

  • A Civitai orchestration token (Quick start → Prerequisites)
  • A training-data zip (signed R2 URL, Civitai R2 AIR, or any HTTPS URL)
  • An accurate count of images in the zip

Chroma ​

Trains on the Chroma1-HD base. Uses TextToImageV2Job for sample renders; output LoRA is usable wherever Chroma is supported.

http
POST https://orchestration.civitai.com/v2/consumer/workflows?wait=0
Authorization: Bearer <your-token>
Content-Type: application/json

{
  "tags": ["training"],
  "steps": [{
    "$type": "training",
    "priority": "normal",
    "retries": 2,
    "input": {
      "engine": "ai-toolkit",
      "ecosystem": "chroma",
      "steps": 2000,
      "resolution": 1024,
      "lr": 0.0001,
      "trainTextEncoder": false,
      "lrScheduler": "cosine",
      "optimizerType": "adamw8bit",
      "networkDim": 16,
      "networkAlpha": 16,
      "trainingData": {
        "type": "zip",
        "sourceUrl": "https://civitai-delivery-worker-prod.5ac0637cfd0766c97916cefa3764fbdf.r2.cloudflarestorage.com/training-images/5418/2382561TrainingData.B6Tr.zip",
        "count": 10
      },
      "samples": {
        "prompts": [
          "woman with red hair, playing chess at the park, dramatic explosion in background",
          "a woman holding a coffee cup, in a beanie, sitting at a cafe",
          "a horse acting as a DJ at a night club, fisheye lens, smoke machine, laser lights"
        ]
      }
    }
  }]
}
POST/v2/consumer/workflows
Set your Civitai API token via the Token button in the navbar to enable Try It.
Request body — edit to customize (e.g. swap the image URL or prompt)
Valid JSON

Chroma defaults: networkDim: 16, optimizerType: adamw8bit, trainTextEncoder: false, lrScheduler: cosine. Default price 2000 Buzz.

ERNIE ​

Trains on Baidu's ERNIE-Image. Comfy-based ecosystem with built-in diffuser. Uses ComfyImageGenJob for sample renders.

http
POST https://orchestration.civitai.com/v2/consumer/workflows?wait=0
Authorization: Bearer <your-token>
Content-Type: application/json

{
  "tags": ["training"],
  "steps": [{
    "$type": "training",
    "priority": "normal",
    "retries": 2,
    "input": {
      "engine": "ai-toolkit",
      "ecosystem": "ernie",
      "steps": 2000,
      "lr": 0.0001,
      "trainTextEncoder": false,
      "lrScheduler": "cosine",
      "optimizerType": "adamw8bit",
      "networkDim": 32,
      "networkAlpha": 32,
      "trainingData": {
        "type": "zip",
        "sourceUrl": "urn:air:other:other:civitai-r2:civitai-delivery-worker-prod@training-images/7918795/2435272TrainingData.bJ7P.zip",
        "count": 10
      },
      "samples": {
        "prompts": ["a portrait of TOK", "TOK walking through a comic book city"]
      }
    }
  }]
}
POST/v2/consumer/workflows
Set your Civitai API token via the Token button in the navbar to enable Try It.
Request body — edit to customize (e.g. swap the image URL or prompt)
Valid JSON

ERNIE defaults: networkDim: 32, optimizerType: adamw8bit, trainTextEncoder: false, lrScheduler: cosine. Default price 1000 Buzz.

Qwen ​

Trains on Qwen-Image. The version field selects a specific Qwen-Image release:

versionBase resolved to
latest (default)Qwen/Qwen-Image-Edit-2512
2509urn:air:qwen:checkpoint:civitai:1864281@2110043
2512Qwen/Qwen-Image-Edit-2512 (same as latest)
http
POST https://orchestration.civitai.com/v2/consumer/workflows?wait=0
Authorization: Bearer <your-token>
Content-Type: application/json

{
  "tags": ["training"],
  "steps": [{
    "$type": "training",
    "priority": "normal",
    "retries": 2,
    "input": {
      "engine": "ai-toolkit",
      "ecosystem": "qwen",
      "version": "latest",
      "steps": 2000,
      "resolution": 1024,
      "lr": 0.00011,
      "trainTextEncoder": false,
      "lrScheduler": "cosine",
      "optimizerType": "adamw8bit",
      "networkDim": 16,
      "networkAlpha": 16,
      "trainingData": {
        "type": "zip",
        "sourceUrl": "urn:air:other:other:civitai-r2:civitai-delivery-worker-prod@training-images/3315022/2526079TrainingData.o4S8.zip",
        "count": 10
      },
      "samples": {
        "prompts": [
          "woman with red hair, playing chess at the park, dramatic explosion in background",
          "a woman holding a coffee cup, in a beanie, sitting at a cafe"
        ]
      }
    }
  }]
}
POST/v2/consumer/workflows
Set your Civitai API token via the Token button in the navbar to enable Try It.
Request body — edit to customize (e.g. swap the image URL or prompt)
Valid JSON

Qwen defaults: networkDim: 16, optimizerType: adamw8bit, trainTextEncoder: false, lrScheduler: cosine. Default price 2000 Buzz.

Z-Image Turbo ​

Trains on ostris/Z-Image-De-Turbo and pulls in the original Tongyi-MAI/Z-Image-Turbo as an extras model. Output LoRA is usable in Z-Image generation on the turbo model.

http
POST https://orchestration.civitai.com/v2/consumer/workflows?wait=0
Authorization: Bearer <your-token>
Content-Type: application/json

{
  "tags": ["training"],
  "steps": [{
    "$type": "training",
    "priority": "normal",
    "retries": 2,
    "input": {
      "engine": "ai-toolkit",
      "ecosystem": "zimageturbo",
      "steps": 2000,
      "resolution": 512,
      "lr": 0.000611,
      "trainTextEncoder": false,
      "lrScheduler": "cosine",
      "optimizerType": "adamw8bit",
      "networkDim": 32,
      "networkAlpha": 32,
      "trainingData": {
        "type": "zip",
        "sourceUrl": "urn:air:other:other:civitai-r2:civitai-delivery-worker-prod@training-images/3315022/2526079TrainingData.o4S8.zip",
        "count": 10
      },
      "samples": {
        "prompts": ["a photo of TOK", "TOK in a garden", "TOK portrait"]
      }
    }
  }]
}
POST/v2/consumer/workflows
Set your Civitai API token via the Token button in the navbar to enable Try It.
Request body — edit to customize (e.g. swap the image URL or prompt)
Valid JSON

Z-Image Turbo defaults: networkDim: 32, optimizerType: adamw8bit, trainTextEncoder: false. Default price 1000 Buzz.

Z-Image Base ​

Trains on Tongyi-MAI/Z-Image. The orchestrator overrides optimizerType to automagic and lr to 0.000001 regardless of what you submit — the input fields are accepted but ignored. Use the Z-Image Turbo recipe instead unless you specifically need a base-model LoRA.

http
POST https://orchestration.civitai.com/v2/consumer/workflows?wait=0
Authorization: Bearer <your-token>
Content-Type: application/json

{
  "tags": ["training"],
  "steps": [{
    "$type": "training",
    "priority": "normal",
    "retries": 2,
    "input": {
      "engine": "ai-toolkit",
      "ecosystem": "zimagebase",
      "steps": 2000,
      "resolution": 512,
      "lr": 0.000611,
      "trainTextEncoder": false,
      "lrScheduler": "cosine",
      "networkDim": 32,
      "networkAlpha": 32,
      "trainingData": {
        "type": "zip",
        "sourceUrl": "urn:air:other:other:civitai-r2:civitai-delivery-worker-prod@training-images/3315022/2526079TrainingData.o4S8.zip",
        "count": 10
      },
      "samples": {
        "prompts": ["a photo of TOK", "TOK in a garden", "TOK portrait"]
      }
    }
  }]
}
POST/v2/consumer/workflows
Set your Civitai API token via the Token button in the navbar to enable Try It.
Request body — edit to customize (e.g. swap the image URL or prompt)
Valid JSON

Z-Image Base defaults: networkDim: 32, optimizerType: automagic (overridden), lr: 0.000001 (overridden), trainTextEncoder: false. Default price 1000 Buzz.

Common parameters ​

Defaults shown are the post-ApplyDefaults values; per-ecosystem deviations are noted above.

FieldRequiredDefaultNotes
engine✅—Always ai-toolkit.
ecosystem✅—One of: chroma, ernie, qwen, zimageturbo, zimagebase.
version(qwen only)latestlatest, 2509, 2512. Selects the Qwen-Image base release.
steps20001–10000. Total training steps. Primary driver of training length and pricing.
epochs101–20. Number of saved checkpoints delivered, each separately downloadable. Each adds 10 Buzz of storage.
batchSize1Defaults to 1. For Z-Image and ERNIE, raise it up to the ecosystem maximum (2) to train faster at the cost of more GPU memory; a larger batch needs fewer steps. For Chroma and Qwen it is fixed at 1. Values above the max are clamped.
continueFrom(none)A previously-trained LoRA AIR (urn:air:<ecosystem>:lora:...) to resume from (see Continue training). Must be a LoRA of the same ecosystem.
lr0.0001UNet learning rate.
trainTextEncoderfalseAll five ecosystems leave the text encoder frozen.
lrSchedulercosineconstant, constant_with_warmup, cosine, linear, step.
optimizerTypeadamw8bit (automagic for Z-Image Base)Full enum on the SDXL/SD1 page.
networkDim32 (16 for Chroma / Qwen)1–256.
networkAlphamatches networkDim1–256.
noiseOffset00–1.
flipAugmentationfalseRandom horizontal flips.
shuffleTokens / keepTokensfalse / 0Caption-tag shuffling.
triggerWord(none)Activation token. Recommended for character / style LoRAs on Chroma, Z-Image.
trainingData.{type, sourceUrl, count}✅—type: "zip".
samples.prompts[][]Preview prompts rendered at each saved checkpoint with the trained LoRA.
samples.negativePrompt(none)—
samples.cfgScale(ecosystem default)Overrides the CFG / guidance scale used when rendering the preview samples.
samples.strength1.0Trained-LoRA weight applied in the preview samples.

Continue training / train further ​

To resume from a LoRA you already trained instead of starting from the base checkpoint, set continueFrom to that LoRA's AIR. The new run starts from those weights and the new epochs build on top:

json
{
  "$type": "training",
  "input": {
    "engine": "ai-toolkit",
    "ecosystem": "chroma",
    "continueFrom": "urn:air:chroma:lora:civitai:<id>@<version>",
    "steps": 1000
  }
}

continueFrom must point at a LoRA of the same ecosystem as the model being trained (a chroma LoRA for ecosystem: "chroma", a qwen LoRA for ecosystem: "qwen", and so on) — a mismatched ecosystem is rejected.

Reading the result ​

Same envelope as the other training recipes — see SDXL/SD1 → Reading the result. Each saved checkpoint yields a .safetensors LoRA blob plus any sample images.

The trained LoRA is usable in the corresponding generation recipe — Chroma LoRAs in any Chroma workflow, ERNIE LoRAs in ERNIE image generation, Qwen LoRAs in Qwen image generation, Z-Image LoRAs in Z-Image generation.

Runtime ​

Per-step wall time, default settings on a 10-image dataset:

EcosystemPer-stepTypical full run
chroma~0.6–1.2 s10–30 min for 2000 steps
ernie~0.3–0.6 s6–16 min for 2000 steps
qwen~0.6–1.2 s10–30 min for 2000 steps
zimageturbo~0.1–0.25 s2–8 min for 2000 steps
zimagebase~0.1–0.25 s2–8 min for 2000 steps

Always use wait=0.

Cost ​

Training is billed per step plus a flat per-epoch storage surcharge, with a price floor:

price = steps × costPerStep + epochs × 10        (rounded)
floor: never less than 80% of the default-configuration price

epochs is the number of saved checkpoints delivered (default 10, range 1–20); each adds 10 Buzz of storage. The default run is 2000 steps / 10 epochs, which lands each ecosystem on its default price. Lowering steps or epochs can save at most 20% (the floor).

EcosystemcostPerStepDefault price (2000 steps, 10 epochs)Floor (80%)
chroma0.9520001600
ernie0.451000800
qwen0.9520001600
zimageturbo0.451000800
zimagebase0.451000800

Sample-prompt rendering is billed separately at each ecosystem's image-generation rate. Use whatif=true (the Preview cost button on the widgets above) to confirm exact charges before submitting.

Troubleshooting ​

SymptomLikely causeFix
400 with "ecosystem unknown"Typo, or not one of chroma / ernie / qwen / zimageturbo / zimagebaseCheck spelling.
400 with "version not allowed" (Qwen only)version not one of latest / 2509 / 2512Use one of the listed values.
Z-Image Base: optimizerType you set seems ignoredIntentional — ApplyDefaults overrides to automagicUse Z-Image Turbo if you need full optimizer control.
Trained LoRA underbakedToo few steps / too low lrRaise steps (these ecosystems often need more steps than SDXL); keep lr ≤ 5e-4.
Trained LoRA overcookedToo many steps or networkDim too highDrop networkDim to 16, lower steps.
Step failed, moderationStatus: "Rejected"Dataset failed content moderationReplace flagged images.

Civitai Developer Documentation

© 2026 Civitai