Todos los modelos de IA en FlyAIgh
38 familias de modelos con los principales generadores de vídeo e imagen con IA. Cada plan de pago incluye todos los modelos: elige por capacidad, no por nivel.
Modelos de video
20 familiasSora
por OpenAI
OpenAI's flagship text-to-video model. Cinematic realism, native audio, 4–12s clips up to 1080p.
VEO 3.1
por Google
Google DeepMind's VEO 3.1 — three tiers of 8-second video with native audio, image-to-video and first/last frame control.
Gemini Omni
por Google
Google's Gemini Omni 1.1 Flash — 3 to 10 second clips from text, an image, or reference images and video clips, with sound included, from 360p up to native 4K.
Seedance 2.5
por ByteDance
ByteDance's newest flagship — single continuous takes up to 30 seconds, native synchronized audio, and up to 30 reference images in one generation.
Seedance 2.0
por ByteDance
ByteDance's production video family — cinematic camera work, native audio, multimodal video-to-video references, and the only native 4K render on FlyAIgh.
Kling V4
por Kuaishou
Kuaishou's newest Kling — single takes from 3 to 30 seconds with stereo sound, first/last-frame control, and a reference mode that takes up to 10 images and 5 reference videos.
Kling V3 Omni
por Kuaishou
Kuaishou's Kling V3 Omni — 3 to 15 second clips from text, a first frame or up to 7 reference images, with a sound switch and a 4K tier, from 8 credits a second.
Kling V3
por Kuaishou
Kuaishou's Kling V3 — 3 to 15 second clips from text, a first frame or a first and last frame, at 720p or 1080p with a sound switch, from 8 credits a second.
FLUX 3 Video
por Black Forest Labs
Keyframe-driven video from Black Forest Labs — pin up to 10 stills on a timeline and the model animates between them, with synced audio and lip-sync.
MiniMax H3
por MiniMax
480p to 1080p from 2 credits a second, or native 2K with stereo audio — 5–15 second clips, first/last-frame control, and multimodal image, video, and audio references.
MiniMax Hailuo 2.3
por MiniMax
Instruction-focused text and image animation with 15 named camera commands, 768p/1080p output, and lower-cost Standard and Fast variants.
Grok Imagine
por xAI
xAI's video model — the cheapest flagship on FlyAIgh at 3 credits per second. Natural motion dynamics, native audio, 6 or 10 second clips.
HappyHorse 1.1
por Alibaba
Alibaba's HappyHorse 1.1 — 3 to 15 second clips that pick text, image or reference mode from what you upload, with up to 9 reference images, at a 40% launch discount.
HappyHorse 1.0
por Alibaba
Alibaba's original HappyHorse — 3 to 15 second clips that switch between text, image and reference mode on their own, with up to 9 reference images at 720p or 1080p.
WAN 3.0
por Alibaba
Alibaba's long-take video model — up to 30 seconds in a single take with native audio and multimodal image, video, and audio references.
WAN 2.7 Video
por Alibaba
Alibaba's WAN 2.7 video model — multi-shot narrative generation with full mode coverage (t2v / i2v / r2v / edit). Up to 1080p, 2–15s clips.
Vidu Q4 Preview
por ShengShu
ShengShu's Vidu Q4 Preview — animate one image or compose up to 12 reference images into 3–16 second clips from 540p to 4K, with sound included.
Vidu Q3 Pro
por ShengShu
ShengShu's Vidu Q3 Pro — text, a first frame, a first and last frame, or up to 7 reference images, in fixed 4, 5 or 8 second clips at 720p or 1080p.
Pixverse C1
por Pixverse
Pixverse C1 — the action and effects specialist for explosions, transformations, vehicles and sport, in 5 or 8 second clips at 720p or 1080p from 6 credits a second.
Pixverse v6
por Pixverse
Pixverse v6 — Pixverse's general-purpose video model for people, products, places and conversation, in 5 or 8 second clips at 720p or 1080p from 6 credits a second.
Modelos de imagen
18 familiasNano Banana 2.1
por Google
Google's Nano Banana 2.1 — sharper detail and edits that hold up over many rounds, at 1K, 2K or 4K, with up to 14 reference images and 14 aspect ratios, from 2 credits an image.
Nano Banana Pro
por Google
Google's Nano Banana Pro — the flagship Nano Banana for finished images: 4K output, image editing with up to 8 references, and strong text rendering and fine detail.
Nano Banana 2
por Google
Google's Nano Banana 2 — fast photorealistic text-to-image at 1K, 2K or 4K in five aspect ratios, for social posts, blog illustrations and prompt iteration.
GPT Image 2.6
por OpenAI
OpenAI's newest image model in two lines — Sunburst for detail that holds across edits, Flare for speed — with exact text, up to 4K, 16 reference images and a free Complexity control.
GPT Image 2.5
por OpenAI
OpenAI's GPT Image 2.5 in two lines — Sunburst for fine detail and exact text, Flare for fast everyday edits — with up to 4K, 15 aspect ratios, 16 references and transparent background.
GPT Image 2
por OpenAI
OpenAI's GPT Image 2 — the earlier GPT Image generation, with clean in-image text, editing from up to 4 references, three quality tiers and 1K to 4K output.
WAN 2.7 Pro
por Alibaba
Alibaba's professional image model — 4K output with Thinking Mode for prompt analysis, watermark control and up to 9 reference images.
Midjourney v8.2 Preview
por Midjourney
Midjourney v8.2 Preview — an early look at Midjourney's next aesthetic step, with richer light, cleaner texture and steadier characters, in seven aspect ratios without Discord.
Midjourney v8.1
por Midjourney
Midjourney v8.1 — the stable Midjourney v8 model, with sharper detail and stronger prompt adherence than v7, in seven aspect ratios without Discord.
Midjourney v7
por Midjourney
Midjourney v7 — Midjourney's established aesthetic with Draft, Fast and Turbo at one flat price, seven aspect ratios and image reference, without Discord.
Seedream 5.5
por ByteDance
ByteDance's Seedream 5.5 in two lines at one price — photoreal images at 1K or 2K, editing with up to 10 reference images, and a native transparent background straight from a prompt.
Seedream 5.0 Pro
por ByteDance
ByteDance's Seedream 5.0 Pro in two lines — Pro for production-grade realism and precise edits, Pro-N for structure-locked redraws — with up to 6 references and ultrawide 21:9.
Seedream 5.0 Lite
por ByteDance
ByteDance's Seedream 5.0 Lite — photorealistic text-to-image at 2K or 3K with ultrawide 21:9 and ultratall 9:21 ratios, from 3 credits an image.
Z Image Turbo
por Z.ai
Z.ai's Z Image Turbo — the cheapest image model on FlyAIgh at just 1 credit per image. Fast turnaround, 1K or 1.5K output.
Qwen Image 2.0 Pro
por Alibaba
Alibaba's Qwen Image 2.0 Pro — the higher-fidelity Qwen, with clean Chinese and English text, finer detail and text-to-image plus editing in one model.
Qwen Image 2.0
por Alibaba
Alibaba's Qwen Image 2.0 — clean Chinese and English text inside images, with text-to-image and editing in one model, from 2 credits an image.
Kling Image 3.0 Omni
por Kuaishou
Kuaishou's Kling Image 3.0 Omni — compose up to 7 reference images into a new still, built for keeping the same character's face, outfit and styling across shots.
Kling Image 3.0
por Kuaishou
Kuaishou's Kling Image 3.0 — text-to-image tuned for portraits, character art and fashion, with strong subject fidelity, from 4 credits an image.