Skip to main content

Advanced Options

Many AI models expose advanced parameters that let you fine-tune generation quality, speed, and output. These options are available in the node configuration panel in the Flow Builder.

Parameter reference

Tier

Controls the quality/speed tradeoff.
ModelOptionsCost impact
Kling O3, 3.0Standard / ProPro costs 2x
Kling 2.6 MotionStandard / ProPro costs 1.5x
Wan 2.6Standard / FlashFlash is cheaper
Sora 2Standard / ProPro costs ~1.4x
Veo 3.1Standard / FastFast is cheaper
LTX 2.3Standard / FastFast applies 0.7x multiplier
FLUX.1Dev / ProPro costs ~1.5x
Nano Bananav2 / ProPro costs 1.5x

Duration

Video length for Kling models.
OptionCost impact
5 secondsBase price
10 seconds2x base price
Available on: Kling O3, Kling 3.0, Kling 2.6.

Resolution

Output resolution for supported models.
OptionCost impact
720pBase price
1080pBase price
1440p1.5x (LTX 2.3)
4K2x (Sora 2, LTX 2.3)
Available on: Sora 2 (720p/1080p/4K), LTX 2.3 (via cost tiers).

Seed

A fixed integer for reproducible generation. Using the same seed with the same prompt and settings produces the same (or very similar) output. Available on: All Kling models, Wan 2.6, Sora 2, Seedance 1.5, MiniMax Hailuo 2, LTX 2.3, FLUX.1, Nano Banana, Grok Imagine, Seedream v5, Qwen Image 2 Pro.

Guidance Scale

Controls how closely the model follows your prompt. Higher values = more literal prompt adherence but potentially less natural results.
ModelRangeDefault
FLUX.11–203.5
Nano Banana1–203.5
LTX 2.31–103.5

Inference Steps

Number of denoising steps during generation. More steps = higher quality but slower generation.
ModelRangeDefault
FLUX.11–10028
Nano Banana1–10030
LTX 2.31–10030

Sync Mode

When enabled, generation completes synchronously (waits for result). When disabled, generation is queued and polled. Available on: FLUX.1 (default: on).

Number of Images

Generate multiple images in a single request. Available on: FLUX.1 (1–4 images, default: 1). Cost scales linearly with the number of images.

Models with no advanced options

These models have no configurable parameters beyond engine selection:
  • GPT Image 1.5
  • DreamActor v2