Explore a concept
Use free Flux.2 or paid Klein to find a composition. When the idea depends on movement, try a short Seedance Mini draft instead. Compare the same subject before increasing the budget.
AI image and video models
Choose a model for images or video. Compare examples, inputs and credit costs before creating.
Browse models
Image models
Flux.2 Pro
Everyday image creation and reference-based editing.
Flux.2 Max
Detailed imagery for final visual work.
Flux.2 Flex
Typography and controlled image creation.
Flux.2 Klein
Fast drafts and visual variations.
Seedream 5.0 Lite
2K–4K tiers for concepts and reference compositions.
Seedream 5.0 Pro
1K/2K creation and up to 4 references.
Nano Banana 2
Flexible resolutions and reference-based creation.
Nano Banana 2.1
Visual layouts and reference editing in 1K–4K.
GPT Image 2.5
Text-led creation and descriptive image editing.
AI video models
Compare video models by their available inputs, output settings and credit costs.
MiniMax H3
Create videos with MiniMax H3 from text or images, with optional start and end frames to guide the action.
MiniMax H3 Max
Create MiniMax H3 Max videos from text or a start/end-frame pair, with 480P, 768P and refined 1080P output.
MiniMax H3 Max Turbo
Explore short video drafts with MiniMax H3 Max Turbo using text, a starting image or a start/end-frame pair.
Seedance 1.0 Pro Fast
Generate silent video drafts with Seedance 1.0 Pro Fast from text or one starting image, from 2 to 12 seconds.
Seedance 1.5 Pro
Create Seedance 1.5 Pro clips from text or images, with optional end frames and selectable generated audio.
Seedance 2.0
Create Seedance 2.0 videos from text or start/end frames, with selectable audio and output up to 1080P in this workspace.
Seedance 2.0 Fast
Try Seedance 2.0 Fast for 480P or 720P text and image animation, with optional audio and end-frame control.
Seedance 2.0 Mini
Explore video ideas with Seedance 2.0 Mini at 480P or 720P, starting from text or a first and optional last frame.
Seedance 2.5
Build longer Seedance 2.5 scenes from text or start/end frames, choosing a fixed duration from 4 to 30 seconds.
Kling 3.0
Generate Kling 3.0 videos from text or start/end frames, with 720P, 1080P and 4K tiers and an audio option.
Kling 3.0 Turbo
Create short Kling 3.0 Turbo videos from a prompt or one starting image, in 720P or 1080P.
Veo 3.1
Create Veo 3.1 clips from text, a starting image or a first/last-frame pair, with selectable audio and resolution.
Veo 3.1 Fast
Try Veo 3.1 Fast for short text and image-based clips, with audio on/off and optional first/last-frame control.
Gemini Omni Flash
Create Gemini Omni Flash videos from text or a single image, with 720P output and a chosen duration of 3–10 seconds.
Start with the result you need
Choose the workflow first, then a model. These starting points connect image creation, editing and video without treating them as separate projects.
Build a product scene, then animate it
Prepare the subject and setting with Nano Banana or Seedream. Check shape, labels and lighting before bringing the saved image into H3. Add an end frame only when you need to guide the finish.
Design around lettering
Compare Flex and GPT Image for a poster or layout. Check spelling at full size before using the artwork as a starting frame; adding motion does not guarantee text will remain unchanged.
Combine visual references
Choose an image model with enough reference capacity to establish the composition first. For video, the current workflow takes a starting image and, on supported models, an ending image—not the same multi-reference controls.
Make a short scene with sound
Use Seedance 1.5 Pro or Veo when you want an audio switch. Gemini offers a simpler single-image workflow without that switch. Keep dialogue short and listen to the whole result.
Prepare the final output
Compare Max or a resolution-selectable image model for still delivery. For moving scenes, choose a duration and output tier that fit the destination: Seedance 2.5 adds time, while Veo and Kling offer larger output options here.
Compare controls and credits
Image and video jobs use different controls, so compare within the relevant table. Values come from the generator configuration; default prices are not a like-for-like quality benchmark.
Compare image model controls
Compare the modes, reference capacity and default text-to-image credits available here. Reference images, resolution and quality can change the total shown in the generator.
| Model | Modes | Max. references | Output choices | Default text credits |
|---|---|---|---|---|
| Flux.2 | Generate / Edit | 1 | Standard | Free |
| Flux.2 Pro | Generate / Edit / References | 4 | Standard | 17 |
| Flux.2 Max | Generate / Edit / References | 4 | Standard | 38 |
| Flux.2 Flex | Generate / Edit / References | 4 | Standard | 38 |
| Flux.2 Klein | Generate / Edit / References | 4 | Standard | 10 |
| Seedream 5.0 Lite | Generate / Edit / References | 4 | 2K / 3K / 4K | 14 |
| Seedream 5.0 Pro | Generate / Edit / References | 4 | 1K / 2K | 54 |
| Nano Banana 2 | Generate / Edit / References | 4 | 0.5K / 1K / 2K / 4K | 32 |
| Nano Banana 2.1 | Generate / Edit / References | 4 | 1K / 2K / 4K | 24 |
| GPT Image 2.5 | Generate / Edit / References | 4 | Low / Medium / High | 16 |
Compare the options available here
Every listed model supports text and image generation. The table shows each model’s default setup, not an equal-settings benchmark. Open a model page for its complete credit table.
| Model | Available length | End frame | Audio | Default setup | Default credits |
|---|---|---|---|---|---|
| MiniMax H3 | 5 / 10 / 15 s | Yes | No separate switch | 480P · 5 s | 100 |
| MiniMax H3 Max | 5–15 s | Yes | No separate switch | 480P · 5 s | 95 |
| MiniMax H3 Max Turbo | 5–15 s | Yes | No separate switch | 480P · 5 s | 50 |
| Seedance 1.0 Pro Fast | 2–12 s | No | No audio | 480P · 2 s | 15 |
| Seedance 1.5 Pro | 4–12 s | Yes | On / off | 480P · 4 s | 30 |
| Seedance 2.0 | 4–15 s | Yes | On / off | 480P · 4 s | 295 |
| Seedance 2.0 Fast | 4–15 s | Yes | On / off | 480P · 4 s | 235 |
| Seedance 2.0 Mini | 4–15 s | Yes | On / off | 480P · 4 s | 150 |
| Seedance 2.5 | 4–30 s | Yes | On / off | 480P · 4 s | 450 |
| Kling 3.0 | 3–15 s | Yes | On / off | 720P · 3 s | 95 |
| Kling 3.0 Turbo | 3–15 s | No | No separate switch | 720P · 3 s | 125 |
| Veo 3.1 | 4 / 6 / 8 s | Yes | On / off | 720P · 4 s | 300 |
| Veo 3.1 Fast | 4 / 6 / 8 s | Yes | On / off | 720P · 4 s | 150 |
| Gemini Omni Flash | 3–10 s | No | No separate switch | 720P · 3 s | 155 |
Make the choice in three steps
Start with the input you already have
Use text when you are exploring a new scene. Use image editing or references when appearance is already decided. For animation, prepare a clean starting frame and check whether your chosen model supports an ending frame.
Pay for the control you need
Reference capacity, output resolution, image quality, video duration and audio can affect cost. Mini, Fast, Turbo and Pro are model-specific tiers, not a universal quality scale. Open the model page when you need the complete settings and price table.
Compare one change at a time
Keep the prompt and source material consistent. For images, inspect lettering, identity and composition; for video, also inspect motion continuity and sound. Fix a weak starting image before animating it, then increase resolution only when the result works.
Choosing and using a model
Is one model best for everything?
No. Your brief, reference images and intended result matter more than a general ranking.
Can I switch models without leaving the editor?
Yes. The selector shows models compatible with your current mode. Check the settings and credit cost when you switch.
Should I create an image before generating a video?
Use text to video to explore a scene you have not designed yet. If composition, clothing or product appearance matters, prepare an image first, then animate it. Ready images in your creation history have a Make video action; it fills the starting frame without submitting a generation.
Why are the default prices not directly comparable?
Defaults differ in duration and resolution, and optional audio can change the price. Compare the same settings where both models support them. The full model pages show available tiers; provider speed claims are not completion-time guarantees.
When is a higher resolution worth choosing?
Choose the size for your final destination after the motion is working. Review subject consistency, small text and fine detail first: increasing resolution will not repair a weak action sequence. Model pages distinguish native output from refinement or upscaling where applicable.