The AI models

Images

Flux Schnell (1 credit): perfect for fast iteration. GPT Image Mini (2): budget-friendly with image editing. Flux 2 Klein (3): the best everyday quality/price ratio. Recraft V4.1 and Seedream 5 Lite (5): polished design, and Seedream accepts your reference images: the champion of recurring characters. Flux Pro 1.1 and Flux Kontext (6): high-end photorealism, Kontext being dedicated to editing an existing image. Flux 2 Pro (7): the most refined output.

Videos

Kling 2.5 Turbo (~30 cr / 5 s): the best starting point, very natural image-to-video. Veo 3.1 Lite (~34 cr / 8 s): surprisingly affordable, with optional generated audio. Grok Imagine Video (~49) and Seedance 2.0 Fast (~56): fast and versatile, Seedance handling first and last frames. Wan 2.7 (~70): great 1080p quality. MiniMax H3 (~110), Veo 3.1 Fast (~112, ~168 with audio) and Kling V3 (~118): the cinematic tier. Seedance 2.5 (~140): the creative flagship.

Displayed costs adapt automatically to the duration, resolution and audio you pick.

Text

The Text node can write your prompts with the best language models: GPT-5, Claude Sonnet 4.5, Gemini, Grok, DeepSeek, Kimi, Llama… Most cost 1 to 3 credits per generation. Give an instruction (“write a detailed cinematic prompt for…”) and wire the result into your Image and Video nodes.

Choosing well

Always start with the cheapest model to validate the idea, then regenerate with a premium model once you like the composition. It is the most credit-efficient workflow, and the AI assistant can choose for you if you describe your need and budget.