AI models

Premium platform. Cheaper pricing. Every AI video and image model on CraftyPeak AI, with live prices, real features and templates for each one, as an alternative to Higgsfield, Runway, OpenArt and HeyGen.

• Seedance 2.5

  Video model by ByteDance. From 13 credits a second.

  Seedance 2.5 on CraftyPeak AI turns text, images, first and last frames and reference videos into video with native sound. Attach up to 40 references and keep characters, products and style consistent across every shot.

• Seedance 2.0

  Video model by ByteDance. From 9 credits a second.

  Seedance 2.0 on CraftyPeak AI creates cinematic, multi-shot video with native sound from text, images, first and last frames or reference video, from 480p drafts to 4K masters.

• Seedance 2.0 Fast

  Video model by ByteDance. From 5 credits a second.

  Seedance 2.0 Fast on CraftyPeak AI is the quicker, lower-cost Seedance 2.0 tier, with the same inputs, references and native sound, for testing ideas and producing social video in volume.

• Seedance 2.0 Mini

  Video model by ByteDance. From 3 credits a second.

  Seedance 2.0 Mini on CraftyPeak AI is the compact Seedance 2.0 tier: the full Seedance input set and native sound at a small price per second, for drafts, loops and bulk video.

• Seedance 1.5 Pro

  Video model by ByteDance. From 4 credits a second.

  Seedance 1.5 Pro on CraftyPeak AI generates clean video from text, an image or first and last frames, with joint audio you can switch on or off.

• Veo 3.1

  Video model by Google DeepMind. From 26 credits a second.

  Veo 3.1 on CraftyPeak AI creates realistic, live-action style video with natural sound from text, an image or first and last frames, from 720p to 4K.

• Veo 3.1 Fast

  Video model by Google DeepMind. From 11 credits a second.

  Veo 3.1 Fast on CraftyPeak AI is the quicker Veo 3.1 tier, with optional sound and the same input modes, from 720p to 4K.

• Veo 3.1 Lite

  Video model by Google DeepMind. From 4 credits a second.

  Veo 3.1 Lite on CraftyPeak AI brings Google's Veo family to drafts, explainers and volume work at 720p and 1080p, with optional sound.

• Google Omni Flash

  Video model by Google. From 14 credits a second.

  Google Omni Flash (Gemini Omni Flash) on CraftyPeak AI creates video from text and edits existing video from a prompt, with native sound, in landscape or portrait.

• Nano Banana Pro

  Image model by Google DeepMind. From 18 credits an image.

  Nano Banana Pro (Gemini 3 Pro Image) on CraftyPeak AI creates and edits high-fidelity images with accurate on-image text, from 1K to 4K.

• Nano Banana 2

  Image model by Google DeepMind. From 6 credits an image.

  Nano Banana 2 (Gemini 3.1 Flash Image) on CraftyPeak AI generates and edits images quickly, from 480p drafts to 4K finals.

• Nano Banana

  Image model by Google DeepMind. From 6 credits an image.

  Nano Banana (Gemini 2.5 Flash Image) on CraftyPeak AI makes fast image generations and conversational photo edits at 1K.

• Seedream 5.0 Pro

  Image model by ByteDance. From 5 credits an image.

  Seedream 5.0 Pro on CraftyPeak AI creates photoreal scenes, product shots and design layouts, and edits uploaded images, up to 2K.

• Seedream 5.0 Lite

  Image model by ByteDance. From 4 credits an image.

  Seedream 5.0 Lite on CraftyPeak AI generates and edits large, detailed images from 2K to 4K at a small price per image.

• FLUX.2 [dev]

  Image model by Black Forest Labs. From 3 credits an image.

  FLUX.2 [dev] on CraftyPeak AI creates detailed images from long, specific prompts and edits uploaded images, at 480p, 1K and 2K.

• FLUX.2 [klein] 9B

  Image model by Black Forest Labs. From 2 credits an image.

  FLUX.2 [klein] 9B on CraftyPeak AI is a compact, quick FLUX.2 model for generation and editing, with more detail than the 4B model.

• FLUX.2 [klein] 4B

  Image model by Black Forest Labs. From 1 credits an image.

  FLUX.2 [klein] 4B on CraftyPeak AI is the smallest FLUX.2 model: very quick generation and editing at a small price per image.

• FLUX.1 [schnell]

  Image model by Black Forest Labs. From 1 credits an image.

  FLUX.1 [schnell] on CraftyPeak AI renders images from text in a few steps, for instant drafts at a small price.

• SDXL Lightning

  Image model by ByteDance. From 1 credits an image.

  SDXL Lightning on CraftyPeak AI is ByteDance's few-step Stable Diffusion XL, for quick text to image and image to image at 480p, 1K and 2K.

• Stable Diffusion 1.5

  Image model by Runway and Stability AI. From 1 credits an image.

  Stable Diffusion 1.5 on CraftyPeak AI restyles an uploaded picture with a prompt while keeping its structure, at 480p, 1K and 2K.

• Nano Banana 2.1

  Image model by Google DeepMind. Coming soon.

  Nano Banana 2.1 is Google's latest Nano Banana model: up to fourteen reference images, inputs including video and PDF, Google Search grounding, adjustable thinking and new ultra-wide aspect ratios.

• Nano Banana 2 Lite

  Image model by Google DeepMind. Coming soon.

  Nano Banana 2 Lite (Gemini 3.1 Flash Lite Image) is Google's fastest Nano Banana tier, targeting under two seconds per image, with fourteen aspect ratios and quick local edits.

• Kling 4.0

  Video model by Kuaishou. Coming soon.

  Kling 4.0 is Kuaishou's flagship video model: up to thirty seconds in one pass, multi-shot storytelling, keyframes, video editing and two-channel sound with lip sync, guided by up to fifteen references.

• Kling 4.0 Flash

  Video model by Kuaishou. Coming soon.

  Kling 4.0 Flash is the faster, lower-cost tier of Kuaishou's Kling 4.0, with text to video, image to video and Omni Reference for videos of up to twenty seconds.

• Kling 3.0

  Video model by Kuaishou. Coming soon.

  Kling 3.0 is Kuaishou's multi-shot video model: storyboard several shots in one prompt, keep elements consistent through camera moves, and give each character native, lip-synced dialogue.

• Kling 3.0 Omni

  Video model by Kuaishou. Coming soon.

  Kling 3.0 Omni is Kuaishou's multimodal video model: combine images, elements, a reference video and bound voices in one generation, or edit existing footage from a prompt.

• Kling 3.0 Turbo

  Video model by Kuaishou. Coming soon.

  Kling 3.0 Turbo is a faster, lower-cost Kling 3.0 tier with text to video, image to video and audio-video sync, reported by Kling partners in mid-2026.

• Kling O1

  Video model by Kuaishou. Coming soon.

  Kling O1 is Kuaishou's unified video model for generation and editing: restyle, inpaint or transform footage with a sentence, keep characters and props consistent, and extend a scene.

• Kling 2.6

  Video model by Kuaishou. Coming soon.

  Kling 2.6 is Kuaishou's first Kling model with native audio: speech, dialogue, singing, ambience and sound effects generated with the picture, plus motion control and reusable custom voices.

• Kling 2.5 Turbo

  Video model by Kuaishou. Coming soon.

  Kling 2.5 Turbo is Kuaishou's cost-efficient Kling video model with stronger prompt adherence, more dynamic motion and camera movement, and realistic physics.

• Wan 3.0

  Video model by Alibaba. Coming soon.

  Wan 3.0 is Alibaba's all-in-one video model: generate up to thirty seconds from text, frames, or references that include images, videos, audio, documents and web links, with switchable multilingual voice.

• Wan 3.0 Prime

  Video model by Alibaba. Coming soon.

  Wan 3.0 Prime is the speed-optimised version of Alibaba's Wan 3.0, keeping its all-in-one references and sound for quicker turnaround.

• Wan 2.7

  Video model by Alibaba. Coming soon.

  Wan 2.7 is Alibaba's video model that plans shots with Thinking Mode, edits video from instructions, continues scenes and keeps up to five characters consistent, with synced sound.

• Wan 2.6

  Video model by Alibaba. Coming soon.

  Wan 2.6 is Alibaba's multi-shot video model with reference-to-video: use a character video to keep a person's appearance and voice, with multi-person dialogue and synced sound.

• Wan 2.5

  Video model by Alibaba. Coming soon.

  Wan 2.5 is Alibaba's first Wan video model with native audio: text to video and image to video with synced sound and lip sync, or video driven by your own audio.

• Wan 2.2

  Video model by Alibaba. Coming soon.

  Wan 2.2 is Alibaba's open-weight mixture-of-experts video model for text to video, image to video and first and last frame, with directable cinematic lighting, composition and colour.

• Wan 2.1

  Video model by Alibaba. Coming soon.

  Wan 2.1 is Alibaba's open-weight video model with text to video, image to video and first and last frame, able to render Chinese and English text inside the video.

• Wan 2.2 Animate

  Video model by Alibaba. Coming soon.

  Wan 2.2 Animate is Alibaba's character animation model: transfer the motion of a reference video onto your character, or replace the person in a video with your character.

• Wan 2.2 Speech-to-Video

  Video model by Alibaba. Coming soon.

  Wan 2.2 Speech-to-Video (S2V) is Alibaba's audio-driven model: give it an image and an audio track, and it animates the character to speak or sing in sync.

• Wan 2.1 VACE

  Video model by Alibaba. Coming soon.

  Wan 2.1 VACE is Alibaba's all-in-one video creation and editing model: generate from references, extend, repaint or restyle video, all in a single open-weight model.

• HappyHorse 1.1

  Video model by Alibaba. Coming soon.

  HappyHorse 1.1 is Alibaba's updated HappyHorse video model, with more expressive motion, stronger character consistency and instruction following, and better audio-video sync.

• HappyHorse 1.0

  Video model by Alibaba. Coming soon.

  HappyHorse 1.0 is Alibaba's video model that debuted at number one on the Artificial Analysis Video Arena for text to video and image to video, with joint audio-video generation and lip sync in seven languages.

• MiniMax H3

  Video model by MiniMax. Coming soon.

  MiniMax H3 is MiniMax's current video model: text to video, first and last frames, a reference mode with up to twelve images, videos and audio files, video motion transfer and 32 kHz stereo sound.

• Hailuo 2.3

  Video model by MiniMax. Coming soon.

  Hailuo 2.3 is MiniMax's video model for complex body motion and physics, facial micro-expressions, and styles from anime and illustration to ink wash and game CG.

• Hailuo 2.3 Fast

  Video model by MiniMax. Coming soon.

  Hailuo 2.3 Fast is the quicker image-to-video tier of MiniMax's Hailuo 2.3, made to cut batch costs.

• Hailuo 02

  Video model by MiniMax. Coming soon.

  Hailuo 02 is MiniMax's video model built on its NCR architecture, known for extreme physics and strong instruction following, from text or an image.

• Vidu Q4

  Video model by ShengShu. Coming soon.

  Vidu Q4 (preview) is ShengShu's video model for lifelike performances: face, body, emotion and voice in sync, guided by up to fifteen images and three audio files.

• Vidu Q3

  Video model by ShengShu. Coming soon.

  Vidu Q3 (Q3 Pro) is ShengShu's video model that generates up to sixteen seconds of video and audio together, with multi-shot camera control, start and end frames and reference to video.

• Vidu Q2

  Video model by ShengShu. Coming soon.

  Vidu Q2 is ShengShu's video model for text to video, image to video and reference to video, with acting-led performances, correct Chinese text and optional sound.

• Vidu Q3 Turbo

  Video model by ShengShu. Coming soon.

  Vidu Q3 Turbo is the faster tier of ShengShu's Vidu Q3, with the Q3 generation's video and sound for quicker turnaround.

• Vidu Q2 Pro

  Video model by ShengShu. Coming soon.

  Vidu Q2 Pro is the advanced tier of ShengShu's Vidu Q2, with reference to video from up to seven images, or four images and two videos, plus video editing and replacement.

• PixVerse V6

  Video model by PixVerse. Coming soon.

  PixVerse V6 is PixVerse's video model with more accurate camera moves, consistent character emotion across cuts, realistic physical interactions, multilingual on-screen text, native sound and multi-shot.

• PixVerse V5.5

  Video model by PixVerse. Coming soon.

  PixVerse V5.5 is PixVerse's video model with native dialogue and lip sync, sound effects and music, multi-part generation and transitions.

• PixVerse V5.6

  Video model by PixVerse. Coming soon.

  PixVerse V5.6 is the last update of PixVerse V5, with better visuals, multilingual vocals and less warping than V5.5.

• Luma Ray3.2

  Video model by Luma AI. Coming soon.

  Luma Ray3.2 is Luma AI's video model for directed shots: image to video with up to sixteen keyframes, Modify Video with performance tracking of up to eight faces, Reframe, and native 16-bit HDR with EXR export.

• Luma Ray3.14

  Video model by Luma AI. Coming soon.

  Luma Ray3.14 is Luma AI's faster Ray3 update with native full-HD output and better temporal consistency, which Luma says is four times faster and three times cheaper per second than Ray3.

• Luma Ray3

  Video model by Luma AI. Coming soon.

  Luma Ray3 is Luma AI's reasoning video model with native 16-bit HDR and EXR, Draft Mode for quick ideas, keyframes, extend and loop.

• LTX-2.5 Pro

  Video model by Lightricks. Coming soon.

  LTX-2.5 Pro is Lightricks' video model with native audio, multi-shot consistency, automatic duration and high-resolution output, from text, images or audio.

• LTX-2.3 Pro

  Video model by Lightricks. Coming soon.

  LTX-2.3 Pro is Lightricks' video model with a rebuilt VAE for sharper output, native portrait video and cleaner audio, plus audio to video, retake and extend.

• LTX-2.5 Fast

  Video model by Lightricks. Coming soon.

  LTX-2.5 Fast is the speed tier of Lightricks' LTX-2.5, generating longer videos of up to twenty seconds with native audio.

• LTX-2.3 Fast

  Video model by Lightricks. Coming soon.

  LTX-2.3 Fast is the speed tier of Lightricks' LTX-2.3, with its sharper output, native portrait video and audio for quick text-to-video and image-to-video work.

• Kling Image 3.0

  Image model by Kuaishou. Coming soon.

  Kling Image 3.0 is Kuaishou's image model for text to image and reference-based editing, keeping features stable across up to ten reference images.

• Kling Image 3.0 Omni

  Image model by Kuaishou. Coming soon.

  Kling Image 3.0 Omni generates consistent series of images from text or references and handles batch edits such as relighting, multi-view and virtual try-on, at native high resolution.

• Kling Image O1

  Image model by Kuaishou. Coming soon.

  Kling Image O1 is Kuaishou's unified image model for generation and editing: describe an edit, or mark it with a doodle, arrow or selection, using up to ten reference images.

• Wan 2.7 Image Pro

  Image model by Alibaba. Coming soon.

  Wan 2.7 Image Pro is the top tier of Alibaba's Wan 2.7 image model: text to image at its largest size, Thinking Mode, custom colour palettes and text in twelve languages.

• Wan 2.7 Image

  Image model by Alibaba. Coming soon.

  Wan 2.7 Image is Alibaba's image model for generation, editing and consistent image sets of up to twelve, with up to nine references, Thinking Mode and custom colour palettes.

• Wan 2.6 Image

  Image model by Alibaba. Coming soon.

  Wan 2.6 Image is Alibaba's Wan image model for text to image and multi-image editing, with interleaved text-and-image output of up to five images.

• Qwen-Image 3.0

  Image model by Alibaba Qwen. Coming soon.

  Qwen-Image 3.0 is Alibaba's text-to-image and editing model for dense layouts: prompts up to 4,500 tokens, readable text down to about ten pixels in twelve languages, LaTeX formulas and extreme aspect ratios.

• Qwen-Image 2.1

  Image model by Alibaba Qwen. Coming soon.

  Qwen-Image 2.1 is Alibaba's unified model for text to image and editing, with up to ten reference images and native transparent (RGBA) output.

• Qwen-Image 2.0

  Image model by Alibaba Qwen. Coming soon.

  Qwen-Image 2.0 is Alibaba's unified generation and editing model: native high-resolution output, edits from one to three input images and up to six results per request.

• Qwen-Image

  Image model by Alibaba Qwen. Coming soon.

  Qwen-Image is Alibaba's 20-billion-parameter open-weight text-to-image foundation model, known for complex text rendering, especially Chinese.

• Qwen-Image-Edit

  Image model by Alibaba Qwen. Coming soon.

  Qwen-Image-Edit is Alibaba's original open-weight Qwen image editor: change an image with a written instruction, including the text inside it.

• Qwen-Image-Layered

  Image model by Alibaba Qwen. Coming soon.

  Qwen-Image-Layered is Alibaba's open-weight model that decomposes an image into separate transparent layers you can edit one by one.

• Qwen-Image-2512

  Image model by Alibaba Qwen. Coming soon.

  Qwen-Image-2512 is Alibaba's December 2025 Qwen-Image release, with more realistic people, richer textures and better text rendering, released with open weights.

• Qwen-Image-Edit-2509

  Image model by Alibaba Qwen. Coming soon.

  Qwen-Image-Edit-2509 is Alibaba's September 2025 Qwen editor with multi-image editing, text editing with font, colour and material control, and built-in ControlNet.

• Qwen-Image-Edit-2511

  Image model by Alibaba Qwen. Coming soon.

  Qwen-Image-Edit-2511 is Alibaba's December 2025 Qwen editor with stronger character consistency, multi-person group photos, built-in community LoRAs and new-viewpoint generation.

• GPT Image 2.5 Sunburst

  Image model by OpenAI. Coming soon.

  GPT Image 2.5 Sunburst is OpenAI's most capable image model for generation and editing, with new extra-high quality levels and a markup overlay that shows exactly where to edit.

• GPT Image 2.5 Flare

  Image model by OpenAI. Coming soon.

  GPT Image 2.5 Flare is OpenAI's fast, high-quality everyday image model, with the same sizes and editing features as Sunburst at lower latency.

• GPT Image 2

  Image model by OpenAI. Coming soon.

  GPT Image 2 is OpenAI's first image model with reasoning, with multilingual text rendering and custom sizes, for generation and editing with up to sixteen input images.

• GPT Image 1.5

  Image model by OpenAI. Coming soon.

  GPT Image 1.5 is an OpenAI image model with better instruction following and precise editing than GPT Image 1. OpenAI is retiring it on 1 December 2026, with GPT Image 2 and 2.5 as successors.

• FLUX 3 Image

  Image model by Black Forest Labs. Coming soon.

  FLUX 3 Image is Black Forest Labs' newest image model: generation and editing in one endpoint, up to ten references, bounding-box layout control, web grounding and very large custom sizes.

• FLUX 3 Image Klein

  Image model by Black Forest Labs. Coming soon.

  FLUX 3 Image Klein is Black Forest Labs' open-weight FLUX 3 model: 5.6 billion parameters, up to ten references, LoRA training on the base model, and high-resolution output.

• FLUX.2 [max]

  Image model by Black Forest Labs. Coming soon.

  FLUX.2 [max] is the top tier of Black Forest Labs' FLUX.2 family, with the highest quality and editing consistency in the family, strong prompt following and web-search grounding.

• FLUX.2 [pro]

  Image model by Black Forest Labs. Coming soon.

  FLUX.2 [pro] is Black Forest Labs' production-grade FLUX.2 model: high quality at an affordable price, hex colour control, structured prompts and up to eight references.

• FLUX.2 [flex]

  Image model by Black Forest Labs. Coming soon.

  FLUX.2 [flex] is Black Forest Labs' adjustable FLUX.2 model, with control over steps and guidance and strengths in typography and preserving small details.

• FLUX.1 Kontext [max]

  Image model by Black Forest Labs. Coming soon.

  FLUX.1 Kontext [max] is Black Forest Labs' top Kontext model for text to image and instruction-based editing, with strong typography, prompt adherence and consistency.

• FLUX.1 Kontext [pro]

  Image model by Black Forest Labs. Coming soon.

  FLUX.1 Kontext [pro] is Black Forest Labs' fast Kontext model for text to image and instruction-based editing, at around five to six seconds per image.

• FLUX1.1 [pro] Ultra

  Image model by Black Forest Labs. Coming soon.

  FLUX1.1 [pro] Ultra is Black Forest Labs' high-resolution FLUX1.1 model with images up to four megapixels, image prompts and a Raw mode for a natural, candid photo look.

• FLUX1.1 [pro]

  Image model by Black Forest Labs. Coming soon.

  FLUX1.1 [pro] is Black Forest Labs' fast, reliable FLUX1.1 text-to-image model with strong prompt adherence.

• Stable Diffusion 3.5 Large

  Image model by Stability AI. Coming soon.

  Stable Diffusion 3.5 Large is Stability AI's 8.1-billion-parameter model, which Stability describes as the family's strongest for prompt adherence and image quality, including typography.

• Stable Diffusion 3.5 Large Turbo

  Image model by Stability AI. Coming soon.

  Stable Diffusion 3.5 Large Turbo is Stability AI's distilled SD 3.5 Large, generating in four steps for fast results.

• Stable Diffusion 3.5 Medium

  Image model by Stability AI. Coming soon.

  Stable Diffusion 3.5 Medium is Stability AI's 2.5-billion-parameter model that runs in about ten gigabytes of memory, with output from 0.25 to 2 megapixels.

• Seedream 4.5

  Image model by ByteDance. Coming soon.

  Seedream 4.5 is ByteDance's image model with improved typography and small text, stronger multi-image editing consistency, up to ten references and batches of up to six.

• Seedream 4.0

  Image model by ByteDance. Coming soon.

  Seedream 4.0 is ByteDance's unified text-to-image and editing model with multi-image references, batch output, knowledge-driven generation and high-resolution output.

• Ideogram 4.5

  Image model by Ideogram. Coming soon.

  Ideogram 4.5 is Ideogram's model for generation, precise mask-based editing and background replacement, using up to five source images.

• Ideogram 4.0

  Image model by Ideogram. Coming soon.

  Ideogram 4.0 is Ideogram's text-rendering model with bounding-box layout, hex colour palettes, JSON prompts, transparent output and wide aspect ratios.

• Ideogram 3.0

  Image model by Ideogram. Coming soon.

  Ideogram 3.0 is Ideogram's model for generation, remix, inpainting, reframing, background replacement, layers and transparent output, with style and character references.

• Recraft V4.1

  Image model by Recraft. Coming soon.

  Recraft V4.1 is Recraft's design model, improving people, gradients and vector and typography precision over V4, with a Pro tier.

• Recraft V4

  Image model by Recraft. Coming soon.

  Recraft V4 is the Recraft design model for raster images, with design-led composition and colour, legible text and style references.

• Recraft V4 Vector

  Image model by Recraft. Coming soon.

  Recraft V4 Vector generates native, editable SVG vectors with the design-led composition and legible text of Recraft V4.

• Recraft V4 Pro

  Image model by Recraft. Coming soon.

  Recraft V4 Pro is the larger raster tier of Recraft V4, with more detailed output for design and marketing work.

• Recraft V4 Pro Vector

  Image model by Recraft. Coming soon.

  Recraft V4 Pro Vector produces more detailed, native, editable SVG vectors from the Pro tier of Recraft V4.

• MiniMax image-01

  Image model by MiniMax. Coming soon.

  MiniMax image-01 is MiniMax's image model for text to image and image to image with a character subject reference, up to nine images per request and eight aspect ratios.