Model catalog

AI Model Catalog

Compare the models connected to TUBAGEN and choose the right engine for each step of your workflow.

53 models

Last updated: August 28, 2026

AI Model Catalog
Provider	Model / version	Type / Supported workflows	Best for	Strengths	Usage boundaries
Google	Gemini Flash 2.5	Text Text generationImage understanding	Fast drafting, summaries, and routine multimodal tasks.	Responsive text generation with image-reference support.	Less suited than Pro models to long, highly complex reasoning.
Google	Gemini Pro 2.5	Text Text generationImage understanding	Detailed writing, analysis, and complex instructions.	Stronger reasoning and instruction handling with image context.	Typically slower and more credit-intensive than Flash.
Google	Gemini Flash 3	Text Text generationImage understanding	Rapid ideation and high-volume text workflows.	Balances speed with current-generation multimodal understanding.	Prioritizes response speed over maximum analytical depth.
Google	Gemini Pro 3	Text Text generationImage understanding	Complex creative briefs and structured production analysis.	High-capability reasoning across text and visual references.	Heavier jobs can take longer than Flash variants.
Google	Gemini Pro 3.1	Text Text generationImage understanding	The most demanding planning, reasoning, and writing tasks.	Our most capable Gemini option for nuanced instructions.	Best reserved for work where added quality justifies extra time and credits.
Anthropic	Claude Opus 4.8	Text Text generationImage understanding	Demanding reasoning, agentic coding, and professional knowledge work.	Handles complex, long-running tasks and supports tool calling.	Uses more credits than lighter text models.
Anthropic	Claude Fable 5	Text Text generationImage understanding	Long-running coding, vision reasoning, and multi-step knowledge work.	Strong planning and verification across extended tasks.	The connected model does not currently support tool calling or Claude Code.
OpenAI	GPT-5.6 Luna	Text Text generationImage understanding	Fast, high-volume text and multimodal workflows.	Lowest-cost GPT-5.6 tier with configurable reasoning effort.	Prioritizes speed and efficiency over maximum capability.
OpenAI	GPT-5.6 Terra	Text Text generationImage understanding	Balanced everyday coding and professional work.	Combines stronger reasoning with practical cost and latency.	Less capable than Sol on the most demanding tasks.
OpenAI	GPT-5.6 Sol	Text Text generationImage understanding	Frontier reasoning, complex coding, and high-autonomy tasks.	The highest-capability GPT-5.6 tier for difficult work.	Uses more credits and can take longer than Luna or Terra.
Google	Nano Banana 2	Image Text to imageImage to image	Reference-led image generation and iterative visual editing.	Supports generation or editing with up to 14 reference images.	Dense multi-reference prompts can still require iteration.
OpenAI	GPT Image 2	Image Text to imageImage to image	Instruction-accurate image creation, text rendering, and edits.	Flexible generation and editing with broad reference support.	4K output uses more credits than 1K or 2K.
xAI	Grok Imagine Image	Image Text to image	Quick image concepts from straightforward prompts.	Simple text-to-image workflow with useful aspect-ratio choices.	Fewer detailed controls than reference-led image models.
xAI	Grok Imagine Image Edit	Image Image to image	Quick, prompt-guided changes to a single image.	A focused image-editing workflow with simple setup.	Accepts one source image and offers fewer output controls than broader editing models.
Alibaba	Wan Image 2.7	Image Text to imageImage to image	Flexible image generation and editing with multiple references.	Supports up to nine references and wide aspect ratios.	Reference-based jobs are capped below the 4K text-only mode.
Alibaba	Wan Image Pro 2.7	Image Text to imageImage to image	Higher-quality Wan image generation and editing.	Pro quality tier with multi-reference and resolution controls.	Reference-based jobs are capped below the 4K text-only mode.
Alibaba	Qwen Image 3.0	Image Text to imageImage to image	Efficient image generation and editing with multilingual typography.	Supports 1K/2K output and edits with up to 3 source images.	Initial editing support accepts JPG/JPEG, PNG, WebP, BMP, and GIF source files; TIFF is not supported.
Alibaba	Qwen Image Pro 3.0	Image Text to imageImage to image	Higher-fidelity layouts, typography, and reference-led image editing.	Pro-quality 1K/2K output with up to 3 source images for editing.	Initial editing support accepts JPG/JPEG, PNG, WebP, BMP, and GIF source files; TIFF is not supported.
ByteDance	Seedream Lite 5.0	Image Text to imageImage to image	Everyday image generation and reference-led edits.	Offers High and Basic quality modes with up to 14 reference images.	TUBAGEN does not expose separate resolution or output-format controls for this option.
ByteDance	Seedream Pro 5.0	Image Text to imageImage to image	Polished image generation and controlled image-to-image work.	High-quality generation with up to ten input images.	Fixed PNG output and a heavier credit cost than Lite.
Google	Nano Banana Pro	Image Text to imageImage to image	Consistent image edits guided by several references.	Supports up to eight reference images and common aspect ratios.	Fewer reference slots than Nano Banana 2.
Google	Nano Banana	Image Text to image	Simple text-to-image generation in varied aspect ratios.	Straightforward, efficient prompt-to-image workflow.	Does not accept reference-image input in TUBAGEN.
Google	Nano Banana Edit	Image Image to image	Prompt-guided edits that combine several source images.	Accepts up to ten source images and supports common aspect ratios.	Requires an input image and is intended for editing rather than text-only generation.
Alibaba	Z-Image	Image Text to imageImage to image	Fast concepts using a small reference set.	Supports up to three references and a broad set of aspect ratios.	Less suitable for jobs needing large reference collections.
Black Forest Labs	FLUX Pro 2	Image Text to imageImage to image	Detailed commercial visuals and images containing typography.	Strong detail with generation or editing from up to eight references.	Complex layouts and exact text may still need retries.
Topaz Labs	Topaz Image Upscale	Image Image upscale	Upscaling a single image to twice its source resolution.	A dedicated image-enhancement tool that needs no prompt.	Requires one source image and currently uses a fixed 2× scale.
Recraft	Recraft Remove Background	Image Background removal	Quickly removing the background from a single image.	A dedicated, prompt-free background-removal tool.	Only removes backgrounds; it does not provide generative image editing.
Recraft	Recraft Crisp Upscale	Image Image upscale	Improving the clarity and detail of a single image.	A dedicated, prompt-free image-enhancement tool.	Focuses on crispness rather than creative generation or detailed controls.
Google	Veo 3.1	Video Text to videoFirst frame to videoFirst and last frame to video	High-quality video guided by first and last frames.	Strong cinematic motion with up to 4K output in the connected workflow.	Quality mode is slower and uses more credits than Fast.
Google	Veo Fast 3.1	Video Text to videoFirst frame to videoFirst and last frame to video	Rapid first/last-frame video drafts.	Faster iteration while retaining frame-guided generation.	Trades some quality for speed.
Google	Veo Reference 3.1	Video Reference to video	Video guided by the style or content of reference images.	Accepts up to three references and supports up to 4K output.	References guide the result; they do not provide frame-by-frame control.
xAI	Grok Imagine Video	Video Text to videoImage to video	Flexible short text-to-video or image-to-video concepts.	Combines text and image modes with durations up to 30 seconds.	Current output is limited to 480p or 720p.
xAI	Grok Imagine Video 1.5 Preview	Video Image to video	Short experimental videos anchored to a source image.	Flexible duration from one to 15 seconds.	Preview model; requires an image and is limited to 480p or 720p.
Alibaba	Wan Video 2.7	Video Text to videoFirst frame to videoFirst and last frame to videoVideo continuationReference to video	Text, frame-guided, continuation, and multimodal reference video.	One model covers five generation modes and up to 1080p.	Reference-to-video clips have a shorter maximum duration.
Alibaba	Wan Video 3.0	Video Text to videoFirst frame to videoFirst and last frame to videoReference to video	Text, frame-guided, and image, video, or audio reference workflows.	Supports flexible clips from two to 30 seconds with output up to 1080p.	With reference video, input and output duration combined must stay within 30 seconds.
MiniMax	MiniMax H3	Video Text to videoImage to videoReference to video	Reference-rich video with optional image, video, or audio guidance.	Supports text, frame, and reference-to-video workflows with image, video, or audio references, up to 2K.	Input requirements vary by mode and need careful setup.
HappyHorse	HappyHorse Video	Video Text to videoImage to videoReference to video	Text, image, and multi-reference video creation.	Combines several generation modes in one model option.	Configured for 720p or 1080p, clips from 3 to 15 seconds, and up to nine reference images.
HappyHorse	HappyHorse Video Edit	Video Video to video	Editing existing video with prompt and image references.	Supports up to five reference images and 720p or 1080p output.	Requires a source video; credit cost varies with source duration and output resolution.
Kuaishou	Kling Video 2.6	Video Text to videoImage to video	Reliable text-to-video and image-to-video generation.	A focused workflow for common video generation modes.	Offers fewer advanced controls than newer Kling variants.
Kuaishou	Kling 3.0	Video Text to videoImage to videoFirst and last frame to videoMulti-shot	High-fidelity text and first/last-frame video with native audio.	Supports Standard, Pro, and 4K tiers with up to five shots.	Multi-shot prompts are limited to 15 seconds in total.
Kuaishou	Kling Turbo 3.0	Video Text to videoImage to video	Faster text or first-frame video iteration.	Produces three-to-15-second clips at 720p or 1080p.	The current integration does not support last-frame guidance.
Kuaishou	Kling AI Avatar Pro 2.0	Video Audio-driven avatar	Long-form 1080p talking avatars driven by a portrait and voice track.	Produces expressive, multilingual avatar video from one image and one audio file.	Requires a prompt; the driving audio must be no longer than five minutes.
Google	Gemini Omni Video	Video Text to videoImage to videoVideo to video	Prompt-led video generation and edits to existing footage.	Combines creation and editing with reference-image controls.	Configured for 720p, 1080p, or 4K at 4, 6, 8, or 10 seconds, with up to seven images and one video reference.
Alibaba	Wan Video Edit 2.7	Video Video to video	Prompt-guided edits to short source videos.	Supports 720p or 1080p output with an optional reference image.	Source clips must be under ten seconds; credit cost varies with duration and resolution.
Topaz Labs	Topaz Video Upscale	Video Video upscale	Increasing the resolution of existing video.	A dedicated video-enhancement tool with 1×, 2×, and 4× processing.	Requires a source video; credit cost varies with duration and upscale factor.
ByteDance	OmniHuman 1.5	Video Audio-driven avatar	Natural talking-human video generated from one image and one voice track.	Strong lip sync, facial expression, and lifelike motion at 720p or 1080p.	The driving audio must be strictly shorter than 60 seconds.
ByteDance	Seedance 2.0	Video Text to videoFirst frame to videoFirst and last frame to videoMultimodal reference	Cinematic text, frame-guided, and multimodal-reference video.	Broad reference support, native audio options, and output up to 4K.	The full-quality tier uses more time and credits than Fast or Mini.
ByteDance	Seedance Fast 2.0	Video Text to videoFirst frame to videoFirst and last frame to videoMultimodal reference	Fast drafts using text, frames, or multimodal references.	Lower-cost, faster access to the Seedance 2.0 reference modes.	Output is limited to 480p or 720p.
ByteDance	Seedance Mini 2.0	Video Text to videoFirst frame to videoFirst and last frame to videoMultimodal reference	Economical exploration of Seedance multimodal workflows.	Supports text, frame, image, video, and audio reference modes.	Output is limited to 480p or 720p.
ByteDance	Seedance 2.5	Video Text to videoFirst frame to videoFirst and last frame to videoMultimodal reference	Longer text, frame-guided, and multimodal-reference video workflows.	Up to 30-second clips, broad multimodal references, and optional native audio.	Output is limited to MP4 at 480p, 720p, or 1080p.
Microsoft	Azure TTS	Audio Text to speech	Voiceover from written scripts.	Selectable voices and speech speed from 0.5× to 2×.	Designed for text-to-speech, not music or sound effects.
Suno	Suno Music V5	Audio Text to music	Instrumental or vocal music from a prompt.	Creates complete music ideas with an instrumental option.	Exact structure, lyrics, and duration may need prompt iteration.
Suno	Suno Sound V5	Audio Text to sound effect	Prompt-based sound effects and looping ambience.	Focused sound generation with an optional loop mode.	Intended for effects rather than speech or full songs.

Google

Gemini Flash

2.5

Text

Supported workflows: Text generationImage understanding
Best for: Fast drafting, summaries, and routine multimodal tasks.
Strengths: Responsive text generation with image-reference support.
Usage boundaries: Less suited than Pro models to long, highly complex reasoning.

Google

Gemini Pro

2.5

Text

Supported workflows: Text generationImage understanding
Best for: Detailed writing, analysis, and complex instructions.
Strengths: Stronger reasoning and instruction handling with image context.
Usage boundaries: Typically slower and more credit-intensive than Flash.

Google

Gemini Flash

Text

Supported workflows: Text generationImage understanding
Best for: Rapid ideation and high-volume text workflows.
Strengths: Balances speed with current-generation multimodal understanding.
Usage boundaries: Prioritizes response speed over maximum analytical depth.

Google

Gemini Pro

Text

Supported workflows: Text generationImage understanding
Best for: Complex creative briefs and structured production analysis.
Strengths: High-capability reasoning across text and visual references.
Usage boundaries: Heavier jobs can take longer than Flash variants.

Google

Gemini Pro

3.1

Text

Supported workflows: Text generationImage understanding
Best for: The most demanding planning, reasoning, and writing tasks.
Strengths: Our most capable Gemini option for nuanced instructions.
Usage boundaries: Best reserved for work where added quality justifies extra time and credits.

Anthropic

Claude Opus

4.8

Text

Supported workflows: Text generationImage understanding
Best for: Demanding reasoning, agentic coding, and professional knowledge work.
Strengths: Handles complex, long-running tasks and supports tool calling.
Usage boundaries: Uses more credits than lighter text models.

Anthropic

Claude Fable

Text

Supported workflows: Text generationImage understanding
Best for: Long-running coding, vision reasoning, and multi-step knowledge work.
Strengths: Strong planning and verification across extended tasks.
Usage boundaries: The connected model does not currently support tool calling or Claude Code.

OpenAI

GPT-5.6 Luna

Text

Supported workflows: Text generationImage understanding
Best for: Fast, high-volume text and multimodal workflows.
Strengths: Lowest-cost GPT-5.6 tier with configurable reasoning effort.
Usage boundaries: Prioritizes speed and efficiency over maximum capability.

OpenAI

GPT-5.6 Terra

Text

Supported workflows: Text generationImage understanding
Best for: Balanced everyday coding and professional work.
Strengths: Combines stronger reasoning with practical cost and latency.
Usage boundaries: Less capable than Sol on the most demanding tasks.

OpenAI

GPT-5.6 Sol

Text

Supported workflows: Text generationImage understanding
Best for: Frontier reasoning, complex coding, and high-autonomy tasks.
Strengths: The highest-capability GPT-5.6 tier for difficult work.
Usage boundaries: Uses more credits and can take longer than Luna or Terra.

Google

Nano Banana

Image

Supported workflows: Text to imageImage to image
Best for: Reference-led image generation and iterative visual editing.
Strengths: Supports generation or editing with up to 14 reference images.
Usage boundaries: Dense multi-reference prompts can still require iteration.

OpenAI

GPT Image

Image

Supported workflows: Text to imageImage to image
Best for: Instruction-accurate image creation, text rendering, and edits.
Strengths: Flexible generation and editing with broad reference support.
Usage boundaries: 4K output uses more credits than 1K or 2K.

xAI

Grok Imagine Image

Image

Supported workflows: Text to image
Best for: Quick image concepts from straightforward prompts.
Strengths: Simple text-to-image workflow with useful aspect-ratio choices.
Usage boundaries: Fewer detailed controls than reference-led image models.

xAI

Grok Imagine Image Edit

Image

Supported workflows: Image to image
Best for: Quick, prompt-guided changes to a single image.
Strengths: A focused image-editing workflow with simple setup.
Usage boundaries: Accepts one source image and offers fewer output controls than broader editing models.

Alibaba

Wan Image

2.7

Image

Supported workflows: Text to imageImage to image
Best for: Flexible image generation and editing with multiple references.
Strengths: Supports up to nine references and wide aspect ratios.
Usage boundaries: Reference-based jobs are capped below the 4K text-only mode.

Alibaba

Wan Image Pro

2.7

Image

Supported workflows: Text to imageImage to image
Best for: Higher-quality Wan image generation and editing.
Strengths: Pro quality tier with multi-reference and resolution controls.
Usage boundaries: Reference-based jobs are capped below the 4K text-only mode.

Alibaba

Qwen Image

3.0

Image

Supported workflows: Text to imageImage to image
Best for: Efficient image generation and editing with multilingual typography.
Strengths: Supports 1K/2K output and edits with up to 3 source images.
Usage boundaries: Initial editing support accepts JPG/JPEG, PNG, WebP, BMP, and GIF source files; TIFF is not supported.

Alibaba

Qwen Image Pro

3.0

Image

Supported workflows: Text to imageImage to image
Best for: Higher-fidelity layouts, typography, and reference-led image editing.
Strengths: Pro-quality 1K/2K output with up to 3 source images for editing.
Usage boundaries: Initial editing support accepts JPG/JPEG, PNG, WebP, BMP, and GIF source files; TIFF is not supported.

ByteDance

Seedream Lite

5.0

Image

Supported workflows: Text to imageImage to image
Best for: Everyday image generation and reference-led edits.
Strengths: Offers High and Basic quality modes with up to 14 reference images.
Usage boundaries: TUBAGEN does not expose separate resolution or output-format controls for this option.

ByteDance

Seedream Pro

5.0

Image

Supported workflows: Text to imageImage to image
Best for: Polished image generation and controlled image-to-image work.
Strengths: High-quality generation with up to ten input images.
Usage boundaries: Fixed PNG output and a heavier credit cost than Lite.

Google

Nano Banana Pro

Image

Supported workflows: Text to imageImage to image
Best for: Consistent image edits guided by several references.
Strengths: Supports up to eight reference images and common aspect ratios.
Usage boundaries: Fewer reference slots than Nano Banana 2.

Google

Nano Banana

Image

Supported workflows: Text to image
Best for: Simple text-to-image generation in varied aspect ratios.
Strengths: Straightforward, efficient prompt-to-image workflow.
Usage boundaries: Does not accept reference-image input in TUBAGEN.

Google

Nano Banana Edit

Image

Supported workflows: Image to image
Best for: Prompt-guided edits that combine several source images.
Strengths: Accepts up to ten source images and supports common aspect ratios.
Usage boundaries: Requires an input image and is intended for editing rather than text-only generation.

Alibaba

Z-Image

Image

Supported workflows: Text to imageImage to image
Best for: Fast concepts using a small reference set.
Strengths: Supports up to three references and a broad set of aspect ratios.
Usage boundaries: Less suitable for jobs needing large reference collections.

Black Forest Labs

FLUX Pro

Image

Supported workflows: Text to imageImage to image
Best for: Detailed commercial visuals and images containing typography.
Strengths: Strong detail with generation or editing from up to eight references.
Usage boundaries: Complex layouts and exact text may still need retries.

Topaz Labs

Topaz Image Upscale

Image

Supported workflows: Image upscale
Best for: Upscaling a single image to twice its source resolution.
Strengths: A dedicated image-enhancement tool that needs no prompt.
Usage boundaries: Requires one source image and currently uses a fixed 2× scale.

Recraft

Recraft Remove Background

Image

Supported workflows: Background removal
Best for: Quickly removing the background from a single image.
Strengths: A dedicated, prompt-free background-removal tool.
Usage boundaries: Only removes backgrounds; it does not provide generative image editing.

Recraft

Recraft Crisp Upscale

Image

Supported workflows: Image upscale
Best for: Improving the clarity and detail of a single image.
Strengths: A dedicated, prompt-free image-enhancement tool.
Usage boundaries: Focuses on crispness rather than creative generation or detailed controls.

Google

Veo

3.1

Video

Supported workflows: Text to videoFirst frame to videoFirst and last frame to video
Best for: High-quality video guided by first and last frames.
Strengths: Strong cinematic motion with up to 4K output in the connected workflow.
Usage boundaries: Quality mode is slower and uses more credits than Fast.

Google

Veo Fast

3.1

Video

Supported workflows: Text to videoFirst frame to videoFirst and last frame to video
Best for: Rapid first/last-frame video drafts.
Strengths: Faster iteration while retaining frame-guided generation.
Usage boundaries: Trades some quality for speed.

Google

Veo Reference

3.1

Video

Supported workflows: Reference to video
Best for: Video guided by the style or content of reference images.
Strengths: Accepts up to three references and supports up to 4K output.
Usage boundaries: References guide the result; they do not provide frame-by-frame control.

xAI

Grok Imagine Video

Video

Supported workflows: Text to videoImage to video
Best for: Flexible short text-to-video or image-to-video concepts.
Strengths: Combines text and image modes with durations up to 30 seconds.
Usage boundaries: Current output is limited to 480p or 720p.

xAI

Grok Imagine Video

1.5 Preview

Video

Supported workflows: Image to video
Best for: Short experimental videos anchored to a source image.
Strengths: Flexible duration from one to 15 seconds.
Usage boundaries: Preview model; requires an image and is limited to 480p or 720p.

Alibaba

Wan Video

2.7

Video

Supported workflows: Text to videoFirst frame to videoFirst and last frame to videoVideo continuationReference to video
Best for: Text, frame-guided, continuation, and multimodal reference video.
Strengths: One model covers five generation modes and up to 1080p.
Usage boundaries: Reference-to-video clips have a shorter maximum duration.

Alibaba

Wan Video

3.0

Video

Supported workflows: Text to videoFirst frame to videoFirst and last frame to videoReference to video
Best for: Text, frame-guided, and image, video, or audio reference workflows.
Strengths: Supports flexible clips from two to 30 seconds with output up to 1080p.
Usage boundaries: With reference video, input and output duration combined must stay within 30 seconds.

MiniMax

Video

Supported workflows: Text to videoImage to videoReference to video
Best for: Reference-rich video with optional image, video, or audio guidance.
Strengths: Supports text, frame, and reference-to-video workflows with image, video, or audio references, up to 2K.
Usage boundaries: Input requirements vary by mode and need careful setup.

HappyHorse

HappyHorse Video

Video

Supported workflows: Text to videoImage to videoReference to video
Best for: Text, image, and multi-reference video creation.
Strengths: Combines several generation modes in one model option.
Usage boundaries: Configured for 720p or 1080p, clips from 3 to 15 seconds, and up to nine reference images.

HappyHorse

HappyHorse Video Edit

Video

Supported workflows: Video to video
Best for: Editing existing video with prompt and image references.
Strengths: Supports up to five reference images and 720p or 1080p output.
Usage boundaries: Requires a source video; credit cost varies with source duration and output resolution.

Kuaishou

Kling Video

2.6

Video

Supported workflows: Text to videoImage to video
Best for: Reliable text-to-video and image-to-video generation.
Strengths: A focused workflow for common video generation modes.
Usage boundaries: Offers fewer advanced controls than newer Kling variants.

Kuaishou

Kling

3.0

Video

Supported workflows: Text to videoImage to videoFirst and last frame to videoMulti-shot
Best for: High-fidelity text and first/last-frame video with native audio.
Strengths: Supports Standard, Pro, and 4K tiers with up to five shots.
Usage boundaries: Multi-shot prompts are limited to 15 seconds in total.

Kuaishou

Kling Turbo

3.0

Video

Supported workflows: Text to videoImage to video
Best for: Faster text or first-frame video iteration.
Strengths: Produces three-to-15-second clips at 720p or 1080p.
Usage boundaries: The current integration does not support last-frame guidance.

Kuaishou

Kling AI Avatar Pro

2.0

Video

Supported workflows: Audio-driven avatar
Best for: Long-form 1080p talking avatars driven by a portrait and voice track.
Strengths: Produces expressive, multilingual avatar video from one image and one audio file.
Usage boundaries: Requires a prompt; the driving audio must be no longer than five minutes.

Google

Gemini Omni Video

Video

Supported workflows: Text to videoImage to videoVideo to video
Best for: Prompt-led video generation and edits to existing footage.
Strengths: Combines creation and editing with reference-image controls.
Usage boundaries: Configured for 720p, 1080p, or 4K at 4, 6, 8, or 10 seconds, with up to seven images and one video reference.

Alibaba

Wan Video Edit

2.7

Video

Supported workflows: Video to video
Best for: Prompt-guided edits to short source videos.
Strengths: Supports 720p or 1080p output with an optional reference image.
Usage boundaries: Source clips must be under ten seconds; credit cost varies with duration and resolution.

Topaz Labs

Topaz Video Upscale

Video

Supported workflows: Video upscale
Best for: Increasing the resolution of existing video.
Strengths: A dedicated video-enhancement tool with 1×, 2×, and 4× processing.
Usage boundaries: Requires a source video; credit cost varies with duration and upscale factor.

ByteDance

OmniHuman

1.5

Video

Supported workflows: Audio-driven avatar
Best for: Natural talking-human video generated from one image and one voice track.
Strengths: Strong lip sync, facial expression, and lifelike motion at 720p or 1080p.
Usage boundaries: The driving audio must be strictly shorter than 60 seconds.

ByteDance

Seedance

2.0

Video

Supported workflows: Text to videoFirst frame to videoFirst and last frame to videoMultimodal reference
Best for: Cinematic text, frame-guided, and multimodal-reference video.
Strengths: Broad reference support, native audio options, and output up to 4K.
Usage boundaries: The full-quality tier uses more time and credits than Fast or Mini.

ByteDance

Seedance Fast

2.0

Video

Supported workflows: Text to videoFirst frame to videoFirst and last frame to videoMultimodal reference
Best for: Fast drafts using text, frames, or multimodal references.
Strengths: Lower-cost, faster access to the Seedance 2.0 reference modes.
Usage boundaries: Output is limited to 480p or 720p.

ByteDance

Seedance Mini

2.0

Video

Supported workflows: Text to videoFirst frame to videoFirst and last frame to videoMultimodal reference
Best for: Economical exploration of Seedance multimodal workflows.
Strengths: Supports text, frame, image, video, and audio reference modes.
Usage boundaries: Output is limited to 480p or 720p.

ByteDance

Seedance

2.5

Video

Supported workflows: Text to videoFirst frame to videoFirst and last frame to videoMultimodal reference
Best for: Longer text, frame-guided, and multimodal-reference video workflows.
Strengths: Up to 30-second clips, broad multimodal references, and optional native audio.
Usage boundaries: Output is limited to MP4 at 480p, 720p, or 1080p.

Microsoft

Azure TTS

Audio

Supported workflows: Text to speech
Best for: Voiceover from written scripts.
Strengths: Selectable voices and speech speed from 0.5× to 2×.
Usage boundaries: Designed for text-to-speech, not music or sound effects.

Suno

Suno Music

Audio

Supported workflows: Text to music
Best for: Instrumental or vocal music from a prompt.
Strengths: Creates complete music ideas with an instrumental option.
Usage boundaries: Exact structure, lyrics, and duration may need prompt iteration.

Suno

Suno Sound

Audio

Supported workflows: Text to sound effect
Best for: Prompt-based sound effects and looping ambience.
Strengths: Focused sound generation with an optional loop mode.
Usage boundaries: Intended for effects rather than speech or full songs.

Credit costs vary with parameters such as resolution, duration, and quality. Refer to the live amount shown in the editor.

This page describes the capabilities currently exposed in TUBAGEN, not every feature offered by each provider. Models, versions, limits, and availability may change.

Put the right model into your workflow

Open TUBAGEN to connect model nodes, compare results, and keep production decisions in one canvas.

Start creating