AAiva Studio

Models

Every image model in Aiva Studio, what it is good for, and typical credit cost per image at default settings (1 image, square).

Open Generate

Platform models

Billed with your Aiva Studio credits (fal.ai and RunPod).

Krea 2

Open weightStandard

Open-weight aesthetic stills — Turbo, Medium, and Large checkpoints.

Text → 9c

Text to image
  • Fast aesthetic generations
  • Turbo / Medium / Large tiers
  • Creative and stylized looks

License: Apache 2.0 — open source (fal safety layer may apply)

Z-Image Turbo

Open weightBudget

Fast text + single-ref variation (not face lock). from 2 credits / image.

Text → 2c

Edit → 2c

Text to imageImage to image · 1 ref max
  • Fast drafts and iterations
  • Single-reference style variations
  • Low-cost exploration

References: 1 ref · style/variation only — not for face lock

License: Apache 2.0 — open source

Flux Kontext Dev

Open weightStandard

Instruction-based edits with steps, guidance, and negative prompt.

Text → 8c

Edit → 8c

Text to imageImage to image · 1 ref max
  • Instruction-based edits
  • Scene and object changes
  • Fine control with steps and guidance

References: 1 ref · instruction-based edits (steps, guidance)

License: Open weights under BFL dev license (not OSI open source)

P-Image Edit

ProprietaryBudget

Style transfer and complex edits — up to 5 reference images.

Text → 3c

Edit → 3c

Text to imageImage to image · 5 refs max
  • Outfit and style transfer
  • Complex multi-reference compositions
  • Creative restyling

References: Up to 5 refs · outfits, style, complex compositions

Qwen Image

Open weightStandard

Open-weight text-to-image with advanced text rendering.

Text → 6c

Edit → 6c

Text to image
  • Text rendering in images
  • Open-weight text-to-image
  • Detailed scene generation

License: Apache 2.0 — open source

Qwen Image LoRA

Open weightStandard

Style-tuned Qwen generation with your own LoRA weights.

Text → 8c

Edit → 8c

Text to image
  • Custom style LoRAs
  • Character or aesthetic tuning
  • Style-specific generation

License: Apache 2.0 — open source

Qwen Image Edit 2511

Open weightStandard

Multi-image edits with strong identity preservation (up to 3 refs).

Text → 6c

Edit → 6c

Text to imageImage to image · 3 refs max
  • Face and identity preservation
  • Multi-reference character edits
  • Consistent likeness across scenes

References: Up to 3 refs · face & identity preservation

License: Apache 2.0 — open source

Qwen Edit 2511 LoRA

Open weightStandard

Multi-ref edits with custom LoRA style transfer.

Text → 8c

Edit → 8c

Text to imageImage to image · 3 refs max
  • LoRA-guided edits
  • Style transfer with refs
  • Anime and custom aesthetics

References: Up to 3 refs · edit with custom LoRA style

License: Apache 2.0 — open source

Seedream 4.0 Edit

ProprietaryStandard

Unified multi-ref editing — outfit swaps, fusion, backgrounds.

Text → 9c

Edit → 9c

Text to imageImage to image · 6 refs max
  • Outfit swaps
  • Scene fusion and backgrounds
  • Many reference images at once

References: Up to 6 refs · outfit swaps, scene fusion

Nano Banana 2

ProprietaryPremium

Google's premium model. Best for character consistency + edits.

Text → 24c

Edit → 24c

Text to imageImage to image · 2 refs max
  • Character consistency
  • Premium text-to-image
  • Quick edits with refs

References: Up to 2 refs · character consistency

Nano Banana Pro

ProprietaryPremium

Google's newest flagship — SOTA quality, typography, and edits.

Text → 45c

Edit → 45c

Text to imageImage to image · 2 refs max
  • Best overall quality
  • Typography in images
  • Flagship character consistency

References: Up to 2 refs · best character consistency

Grok Imagine

ProprietaryStandard

xAI's aesthetic generator. Affordable text + edit.

Text → 6c

Edit → 8c

Text to imageImage to image · 3 refs max
  • Aesthetic, stylized outputs
  • Affordable text and edits
  • Creative looks

References: Up to 3 refs · aesthetic edits

GPT Image 2

ProprietaryPremium

OpenAI's flagship. Best typography & multilingual text.

Text → 99c

Edit → 99c

Text to imageImage to image · 1 ref max
  • Multilingual text in images
  • High-quality typography
  • Prompt-guided single-ref edits

References: 1 ref · prompt-guided edits

Higgsfield (BYOK)

Connect your Higgsfield account in Integrations. Billing runs on your provider account, not credits.

Higgsfield Soul

Text only

Prompt-only influencer-native model. No reference images.

  • Influencer-native prompts
  • BYOK billing
  • No reference images

GPT Image 2

Max quality

Highest quality output, maximum realism.

  • Maximum realism
  • BYOK on your Higgsfield account

Nano Banana 2

Fast

Quick generations with solid quality.

  • Fast quality
  • BYOK

Nano Banana Flash

Fastest

Fastest turnaround for drafts.

  • Fastest drafts
  • BYOK
Connect Higgsfield →