# Supported models

> A working map of video, image and audio models makers use in generative filmmaking, with links out to each provider.

## Video Generation Models

These AI models generate video from text prompts or still images, creating motion sequences for short-film and audiovisual production.

### [FLUX.3](https://bfl.ai/)

Black Forest Labs' video model. Up to twenty seconds in a single generation with native audio, ten keyframes pinned to exact timecodes, and a draft mode you can render up.

Start frame, End frame, Keyframes, Audio, Extend.

### [Veo 3.1](https://deepmind.google/technologies/veo/)

High-quality video generation with native audio from Google DeepMind. Best for final outputs with start and end frame control.

Start frame, End frame, Audio.

### [Veo 3.1 Fast](https://deepmind.google/technologies/veo/)

Faster variant of Veo 3.1 with native audio. Lower cost with quicker generation times.

Start frame, End frame, Audio.

### [Kling 3.0 Pro](https://klingai.com/)

Top-tier video generation with cinematic visuals, fluid motion, native audio, and element support.

Start frame, End frame, Audio, Elements, Multi-shot.

### [Kling O3 Pro](https://klingai.com/)

Video generation with multi-shot support, extended durations up to 15s, and reference-to-video.

Start frame, End frame, Audio, Multi-shot.

### [Kling 2.6](https://klingai.com/)

Top-tier video generation with cinematic visuals, fluid motion, and native audio support.

Start frame, Audio.

### [Kling O1](https://klingai.com/)

High-quality image-to-video generation with smooth motion and end frame control.

Start frame, End frame.

### [MiniMax H3 Max Turbo](https://www.minimax.io/)

A faster, half-price cut of MiniMax H3 Max at about 97% of its quality, with native audio and start/end frames at up to 1080p. Preview.

Start frame, End frame, Audio.

### [MiniMax H3 Max](https://www.minimax.io/)

MiniMax H3 post-trained for stronger prompt adherence and aesthetics, with native audio, start/end frames, and image, video, or audio references at up to 1080p. Extends a clip by up to 15 seconds at up to 2K, replaces a section of one with a new 5-13 second scene, and recasts the people in an existing clip from reference photos.

Start frame, End frame, Audio, Reference media, Extend, Insert scene, Recast.

### [MiniMax H3](https://www.minimax.io/)

Multimodal 2K video generation with native stereo audio and image, video, or audio references.

Start frame, End frame, Audio, Reference media, 2K.

### [Hailuo 02](https://hailuoai.video/)

MiniMax's advanced video generation model with 1080p resolution and start/end frame control.

Start frame, End frame.

### [Seedance 2.5](https://seed.bytedance.com/en/seedance)

ByteDance's latest Seedance model, with native 30-second single takes, an Auto duration option, end frame support, and a 50-file reference budget.

Start frame, End frame, Audio, Reference images, Elements, Auto duration, 30s.

### [Seedance 2.0](https://seed.bytedance.com/en/seedance)

ByteDance's most advanced video generation with cinematic output, native audio, real-world physics, and director-level camera control.

Start frame, End frame, Audio, Reference images, Elements, 4K.

### [Seedance 2.0 Fast](https://seed.bytedance.com/en/seedance)

Faster, more affordable Seedance 2.0 with cinematic output, native audio, and reference-to-video, capped at 720p for quicker, cheaper generations.

Start frame, End frame, Audio, Reference images, Elements.

### [Seedance 2.0 Mini](https://seed.bytedance.com/en/seedance)

The quickest, most affordable native Seedance 2.0 tier with audio and reference media, capped at 720p.

Start frame, End frame, Audio, Reference images, Elements.

### [Seedance 1.5](https://seed.bytedance.com/en/seedance)

High-quality video generation from ByteDance with native audio support and flexible durations from 4-12 seconds.

Start frame, End frame, Audio.

### [Grok Imagine](https://x.ai/)

Video generation from xAI with flexible durations from 1-15 seconds and a wide range of aspect ratios.

Start frame.

### [Grok Imagine 1.5](https://x.ai/)

Text, image, and reference video generation from xAI with native audio, durations up to 15 seconds, and up to 1080p output.

Start frame, References, Audio, 1080p.

### [Grok Imagine 1.5 Lite](https://x.ai/)

Budget text and image video generation from xAI with native audio, durations up to 15 seconds, and 480p to 1080p output.

Start frame, Audio, 1080p.

### [Pixverse V6](https://pixverse.ai/)

High-quality video generation with native audio, flexible durations up to 15 seconds, and art style presets.

Start frame, Audio.

### [LTX 2.3](https://www.lightricks.com/)

Lightricks' open-source 4K video model with native audio. Supports start/end frame control and durations up to 10 seconds.

Start frame, End frame, Audio.

### [Boreal by Creatify](https://fal.ai/models/creatify/boreal)

Product, UGC, and presenter videos with synchronized audio, optional image and driving audio, and durations up to 20 seconds.

Start frame, Audio, Driving audio.

### [Happy Horse 1.0](https://fal.ai/models/alibaba/happy-horse/image-to-video)

Alibaba's #1-ranked video model with native audio, reference-to-video support (up to 9 images), and durations up to 15 seconds.

Start frame, Audio, Reference images.

### [Wan 3.0](https://fal.ai/models/alibaba/wan-3.0/text-to-video)

Alibaba's Wan 3.0 — native 30-second takes at up to 1080p, start and end frames, toggleable native audio, and image, video, or audio references.

Start frame, End frame, Audio, Reference media, Auto duration, 30s.

### [Wan 3.0 Prime](https://fal.ai/models/alibaba/wan-3.0-prime/text-to-video)

Alibaba's accelerated Wan 3.0 tier — faster generation with the same 30-second takes, 1080p output, native audio, and multimodal references.

Faster generation, Start frame, End frame, Audio, Reference media, Auto duration.

### [Luma Ray 3.2](https://fal.ai/models/luma/agent/ray/v3.2/video-to-video)

Luma's video-to-video model — re-render an existing clip into new cinematic motion from a text prompt, with an optional start frame, at up to 1080p.

Video to video, Start frame.

### [Gemini Omni 1.1 Flash](https://deepmind.google/models/gemini/)

Google's Gemini Omni 1.1 Flash — fast text-to-video, image-to-video, and reference-to-video with native synchronized audio and role-bound reference images and clips. Draft at 360p, finish at 4K.

Start + end frame, Reference images, Reference videos, Elements, Audio, Up to 4K.

## Still Generation & Image Editing Models

These models specialise in still image generation and editing — useful for reference frames, concept art, or generating motion from static images.

### [Qwen Image 3.0 Pro](https://fal.ai/models/fal-ai/qwen-image-3/pro/text-to-image)

High-fidelity image generation for dense typography, layouts, and precise edits. Add up to three references to preserve identity and details.

Reference images, Dense typography, Precise editing.

### [Grok Imagine Image 2.0](https://fal.ai/models/xai/grok-imagine-image/v2.0/text-to-image)

xAI's image model for sharp typography and complex layouts, with photorealistic fidelity across photography, design, and illustration. Add up to three references to edit elements and styles.

Reference images, Sharp typography, Photorealistic fidelity.

### [FLUX.3 Image](https://fal.ai/models/blackforestlabs/flux-3/text-to-image)

Black Forest Labs' image model for native 2K and 4K output with precise layout and text. Add up to ten references to make targeted edits that keep everything else intact.

Up to 10 reference images, Native 4K, Targeted edits.

### [Reve 2.1](https://reve.com/)

Native 4K image generation with precise layout planning, spatial control, and multilingual text. Add references to edit or remix.

Reference images, Native 4K, Strong text rendering.

### [Nano Banana 2](https://gemini.google/overview/image-generation/)

Fast, affordable image generation with excellent typography and realism. Supports multi-image blending with character consistency.

Reference images, Web search.

### [Nano Banana 2.1](https://fal.ai/models/fal-ai/nano-banana-2.1)

Google's update to Nano Banana 2 with sharper realism, cleaner text and infographics, and steadier characters across edits.

Up to 14 reference images, Web search, Up to 4K.

### [Nano Banana 2 Lite](https://gemini.google/overview/image-generation/)

Google's fastest, most affordable image model. Fixed 1K output with reference-image blending and natural-language editing.

Reference images.

### [Seedream 5 Lite](https://www.bytedance.com/)

ByteDance's fast, affordable image model. High-resolution 2K–4K output with flexible aspect ratios. Add reference images to edit or blend.

Reference images, 2K–4K output.

### [Seedream 5 Pro](https://www.bytedance.com/)

ByteDance's flagship image model. Sharp high-resolution output with strong text rendering and flexible aspect ratios. Add reference images to edit or blend.

Reference images, Strong text rendering.

### [Nano Banana Pro](https://gemini.google/overview/image-generation/)

State-of-the-art image generation with excellent typography and realism. Supports multi-image blending with character consistency.

Reference images, Web search.

### [Flux 2 Max](https://bfl.ai/)

High-quality image generation with flexible aspect ratios and safety controls. Add reference images for creative editing and blending.

Reference images.

### [GPT Image 2](https://openai.com/)

OpenAI's GPT Image 2 — photoreal image generation with strong typography and instruction following. Add reference images to edit an existing image.

Reference images.

### [GPT Image 2.5 Flare](https://openai.com/)

OpenAI's GPT Image 2.5 Flare — fast, high-quality image generation with natural lighting, rich textures, and precise instruction following. Add reference images to edit an existing image; edits change only what you ask for.

Reference images.

### [GPT Image 2.5 Sunburst](https://openai.com/)

OpenAI's GPT Image 2.5 Sunburst — precision image generation for polished product imagery and campaign creative, with tighter control across edits. Slower than Flare. Add reference images to edit an existing image.

Reference images.

### [Ideogram 4.5](https://fal.ai/models/ideogram/v4.5)

Ideogram's precise image model. Typography renders exactly as written in any language, and edits change only what you ask for, so a run of edits stays clean. Edit with a painted region or guide the change with up to four reference images.

Reference images, Exact typography, Precise editing, Region edit.

## Audio Generation Models

Generate music, sound effects, dialogue, and voiceover. Sonilo SFX can synchronise foley and ambience to an existing shot; VEED Clean Audio and DeepFilterNet 3 clean up noisy dialogue on a clip or from a video source.

### [MiniMax Music](https://www.minimax.io/audio/music)

Generate complete music tracks from a prompt, with optional vocals and custom lyrics.

Music, Instrumental or vocals, Custom lyrics.

### [MiniMax Music 3](https://www.minimax.io/audio/music)

Generate a complete song up to five minutes long, holding theme, vocal identity and arrangement steady across the whole track.

Music, Up to 5 minutes, Custom lyrics.

### [Lyria 3.5](https://deepmind.google/models/lyria/)

Generate complete songs up to about three minutes with richer musicality, lyrics, and multilingual vocals.

Music, Up to 3 minutes, Multilingual vocals.

### [Lyria 3](https://deepmind.google/models/lyria/)

Generate 30-second music sketches with singing vocals and negative-prompt control.

Music, 30 seconds, Vocals.

### [Lyria 3 Pro](https://blog.google/innovation-and-ai/technology/ai/lyria-3-pro/)

Generate complete songs up to about three minutes with vocals, timed lyrics, and full arrangements.

Music, Up to 3 minutes, Timed lyrics.

### [ElevenLabs Music v2.5](https://elevenlabs.io/music)

Generate vocal or instrumental music from three seconds to ten minutes with Eleven Music v2.5, with optional section planning.

Music v2.5, 3 seconds–10 minutes, Section planning.

### [Stable Audio 2.5](https://stableaudio.com/)

Generate stereo music, foley, and ambient sound design up to 190 seconds.

Music and SFX, Up to 190 seconds, Stereo.

### [Seed Audio 1.0](https://seed.bytedance.com/en/models)

Generate dialogue, sound effects, and music together from one prompt, or create clean speech.

Music, SFX, and dialogue, Up to 2 minutes, Voice presets.

### [ElevenLabs SFX v2](https://elevenlabs.io/sound-effects)

Generate short foley, ambience, impacts, and atmospheric sounds from text.

Sound effects, 0.5–22 seconds, Text to SFX.

### [Beatoven SFX](https://www.beatoven.ai/maestro/sound-effects)

Generate stereo effects for creatures, vehicles, impacts, science fiction, and abstract textures.

Sound effects, 1–35 seconds, Stereo.

### [CassetteAI SFX](https://cassetteai.com/)

Generate fast sound-effect placeholders and draft variations up to 30 seconds.

Sound effects, 1–30 seconds, Fast generation.

### [Sonilo SFX](https://fal.ai/models/sonilo/v1.1/text-to-sound-effects)

Generate standalone sound effects from text, or add a source video for frame-synchronised foley and ambience.

Text to sound effects, Video synchronisation, 1–180 seconds, AAC, MP3, WAV, FLAC.

### [DeepFilterNet 3](https://fal.ai/models/fal-ai/deepfilternet3)

Reduce background noise on dialogue and upsample it to 48 kHz. Works on an audio clip or from a video source, and the clean track can replace the audio on the source video.

Noise reduction, 48 kHz upsampling, Audio or video source, Replace audio on source video.

### [VEED Clean Audio](https://fal.ai/models/veed/clean-audio)

Remove background noise from dialogue while keeping quiet and distant words, then level it to broadcast loudness. Works on an audio clip or from a video source, up to 30 minutes.

Noise removal, Keeps quiet words, -19 LUFS loudness, Audio or video source, up to 30 min.

### [ElevenLabs TTS v3](https://elevenlabs.io/v3)

Generate expressive dialogue and voiceover with preset voices, language control, and delivery tags.

Dialogue and voiceover, 21 voices, 70+ languages.

### [ElevenLabs TTS v4](https://fal.ai/models/elevenlabs/tts/eleven-v4)

Generate expressive dialogue and voiceover, directing delivery with inline audio tags and shifting emotion mid-sentence.

Dialogue and voiceover, 21 voices, 80+ languages, Audio tags.

### [ElevenLabs TTS v4 Turbo](https://fal.ai/models/elevenlabs/tts/eleven-v4-turbo)

Generate voiceover fast and at half the price of v4, with the same voices and audio tags.

Dialogue and voiceover, 21 voices, 80+ languages, Low latency.

### [MiniMax Speech-02 HD](https://www.minimax.io/audio/text-to-speech)

Generate natural narration and character dialogue with broad voice and language support.

Dialogue and voiceover, 17 voices, 40+ languages.

### [Gemini 3.8 Flash TTS](https://ai.google.dev/gemini-api/docs/speech-generation)

Generate expressive dialogue and voiceover, directing delivery in plain language: character, accent, pacing, and emotion.

Dialogue and voiceover, 30 voices, Style direction.

### [Gemini 3.8 Flash Lite TTS](https://ai.google.dev/gemini-api/docs/speech-generation)

Generate high-volume voiceover fast and at low cost, with the same voices and style direction as Flash TTS.

Dialogue and voiceover, 30 voices, Style direction.