Wan video LoRA training

Preview ecosystem

Wan video training is currently marked Preview in the orchestrator. The endpoint accepts requests and whatif=true cost previews work, but actual training runs may not be available on every worker fleet. Reach out via Civitai Discord before integrating against production traffic.

Train a WAN video LoRA on a small set of source video clips using AI Toolkit. Output is a video LoRA usable in WAN text-to-video and image-to-video generation.

modelVariant	Wan family	Buzz / epoch
2.1	Wan 2.1 (14B)	12
2.2	Wan 2.2 (14B-A14B)	12

Long-running step

Video training is the slowest training mode on the platform — single-digit minutes per epoch on a 4-clip dataset. Always use wait=0 and follow up via webhook or polling.

The request shape

{
  "$type": "training",
  "input": {
    "engine":       "ai-toolkit",
    "ecosystem":    "wan",
    "modelVariant": "2.1"        // 2.1 | 2.2
  }
}

Prerequisites

• A Civitai orchestration token (Quick start → Prerequisites)
• A training-data zip containing source video clips (each ≤ a few seconds, similar resolution)
• An accurate count of clips in the zip

Wan 2.1 / 2.2

Both variants share the same input shape and per-epoch cost; pick the one that matches your inference target. The example below uses 2.1; swap modelVariant to "2.2" for Wan 2.2 training (no other change required).

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

{
  "tags": ["training", "video"],
  "steps": [{
    "$type": "training",
    "priority": "normal",
    "retries": 2,
    "input": {
      "engine": "ai-toolkit",
      "ecosystem": "wan",
      "modelVariant": "2.1",
      "epochs": 2,
      "resolution": 512,
      "lr": 0.0002,
      "trainTextEncoder": false,
      "lrScheduler": "constant",
      "optimizerType": "adamw8bit",
      "networkDim": 32,
      "networkAlpha": 32,
      "trainingData": {
        "type": "zip",
        "sourceUrl": "urn:air:other:other:civitai-r2:civitai-delivery-worker-prod@training-images/5418/2202966TrainingData.Kjwp.zip",
        "count": 4
      },
      "samples": {
        "prompts": ["a video of TOK", "TOK moving in a garden"]
      }
    }
  }]
}

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)

{ "tags": [ "training", "video" ], "steps": [ { "$type": "training", "priority": "normal", "retries": 2, "input": { "engine": "ai-toolkit", "ecosystem": "wan", "modelVariant": "2.1", "epochs": 2, "resolution": 512, "lr": 0.0002, "trainTextEncoder": false, "lrScheduler": "constant", "optimizerType": "adamw8bit", "networkDim": 32, "networkAlpha": 32, "flipAugmentation": false, "trainingData": { "type": "zip", "sourceUrl": "urn:air:other:other:civitai-r2:civitai-delivery-worker-prod@training-images/5418/2202966TrainingData.Kjwp.zip", "count": 4 }, "samples": { "prompts": [ "a video of TOK", "TOK moving in a garden" ] } } } ] }

Valid JSON

Common parameters

Defaults shown are the post-ApplyDefaults values for Wan.

Field	Required	Default	Notes
engine	✅	—	Always ai-toolkit.
ecosystem	✅	—	Always wan for this page.
modelVariant	✅	—	2.1 or 2.2.
epochs		5	1–20. Billed per epoch. Keep low (2–5) for video — the per-epoch step count is much higher than image.
numberOfRepeats		(no auto-default for Wan)	1–5000.
lr		0.0001	0.0002 is a typical override for video; see example.
trainTextEncoder		false	Leave off — Wan training does not benefit from text-encoder updates.
lrScheduler		cosine	constant, constant_with_warmup, cosine, linear, step.
optimizerType		adamw8bit	See SDXL/SD1 page for full enum.
networkDim		32	1–256.
networkAlpha		matches networkDim	1–256.
noiseOffset		0	0–1.
flipAugmentation		false	Random horizontal flips.
shuffleTokens / keepTokens		false / 0	Caption-tag shuffling.
triggerWord		(none)	Activation token. Per the source, not all video ecosystems support triggerWord — leave empty if you see schema rejections.
trainingData.{type, sourceUrl, count}	✅	—	type: "zip". Zip should contain video clips.
samples.prompts[]		[]	Per-epoch preview videos rendered with the trained LoRA.
samples.negativePrompt		(none)	—

Reading the result

Same envelope as the other training recipes — see SDXL/SD1 → Reading the result. Each epoch yields a video LoRA .safetensors blob plus any sample .mp4 files. The trained LoRA is usable in WAN video generation by referencing it in the loras field.

Runtime

Per-epoch wall time, default settings on a 4-clip dataset:

Variant	Per-epoch	Typical full run
2.1	~3–10 min	6–20 min for 2 epochs
2.2	~3–10 min	6–20 min for 2 epochs

Always use wait=0.

Cost

total = 12 × epochs   (Buzz, base cost)

Cost-per-epoch is 12 per the orchestrator source. Sample-prompt rendering uses Wan video-generation rates (much higher than image samples) and is billed separately. Run with whatif=true to see the exact pre-flight charge.

Configuration	Buzz (training only)
epochs: 2	24 + samples
epochs: 5	60 + samples

Troubleshooting

Symptom	Likely cause	Fix
400 with "modelVariant required"	Missing modelVariant	Set to "2.1" or "2.2".
Step starts then fails immediately	Preview ecosystem not yet enabled on the routing GPU fleet	Contact Civitai support — Wan training is rolling out.
Step failed with VRAM-related error	Resolution × clip length too high for the worker	Lower resolution (e.g. to 512), shorten clips to ≤ 3 seconds.
Trained LoRA produces static / no motion	Too few epochs, too few / too short clips	Raise epochs to 3–5; ensure clips show the motion you want learned.
Step failed, moderationStatus: "Rejected"	Dataset failed content moderation	Replace flagged clips.

Related

• LTX2 video LoRA training — Lightricks LTX video LoRA training (also video, less expensive previews on LTX2.3)
• WAN video generation — use a trained LoRA in WAN inference
• Flux 2 Klein LoRA training — image-side counterpart
• Results & webhooks
• SubmitWorkflow / GetWorkflow
• Endpoint OpenAPI spec