Every model. One bill.

True 4K Kling video in three fixed aspect ratios, with native audio and up to 6-shot sequencing

True 4K output in three aspect ratios with native audio and first/last-frame control, from $0.42/second

LTX-2's faster, lower-latency tier — 4K, 50fps, and optional synced audio

Cinematic LTX-2 Pro text and image to video generator

Omni-reference video generation with first/last frame control

Multi-shot cinematic video generation with native audio, 20+ camera controls, and character consistency

Native synced audio and video in one generation, with output up to 1080p

Premium multimodal video generation with native audio and cinematic motion

Seedance 2.0, tuned for speed — native audio and multi-shot motion intact, minus the flagship's 1080p tier

The lowest-priced tier of Seedance 2.0 — same 720p ceiling as Fast, native audio and multi-shot generation intact

Google's speed-tuned Veo 3.1 model: generate from a prompt, an image, reference photos, or extend an existing clip

The first Wan with reference-to-video and video-to-video editing — four modes, native audio, multi-shot scenes

24fps lip-sync avatar video from a single photo and an audio clip

Aurora's faster variant: turns one photo and one audio clip into a 480p lip-synced avatar video.

Multimodal video generation with native synchronized audio across styles and modes

Multimodal video generation and editing with native audio, references, and multi-turn control

AI video generation with synchronized audio from text and images

Higher-tier Grok video from a start frame or reference images — now up to 1080p

Alibaba's video model with 4 modes — text, image, up to 9 references, or edit an uploaded clip, 720p/1080p, 3-15s

Alibaba's upgraded video model — text, image, or up to 9 reference images, 3-15s clips at 720p/1080p

AI talking avatar video from a HeyGen avatar or your own photo, driven by a script or audio

Script-to-video talking avatars from 500 ready HeyGen looks, no photo needed

One prompt, a finished video: script, avatar, B-roll, and captions, up to 5 minutes long

Kling VIDEO 2.6 Pro is a full audio-visual AI video model that combines cinematic-quality video generation with native audio (dialogue, sound effects, ambience), with optional Motion Control for precise character movement via the API.

High-fidelity multimodal video generation with native audio and advanced editing

Multimodal video generation with native audio and efficient performance

Fast Kling 3.0 video with native audio built in at no extra cost, up to 6 shots per generation

Text-to-video, image-to-video, video editing, and motion transfer in one model — plus native audio.

Kling 3.0's budget tier — 720p video with native audio and prompt-based edits

Turns a start and end frame into a 1080p video clip, 5 or 10 seconds long

Turn a text prompt or a single photo into a steady, accurate 720p video clip

Anchor a first and last frame from your own photos, then render a 5 or 10 second 1080p clip

1080p text- or image-to-video with 3-image character reference — the flagship Kling 2.1 tier

KlingAI 2.1 Pro for cinematic AI video generation

Animate a photo into 1080p video — the budget-friendly Kling 2.1 tier

Kling's higher tier: text and image to video in one model, up to 10-second clips, two guide frames

Kling 2.5 Turbo's budget tier — 720p image-to-video in 5 or 10 second clips

Photo plus audio becomes a 5-minute talking avatar video, no prompt required

Turns a photo and an audio track into a lip-synced avatar video

Redub existing video to new dialogue or music, with independent original and synced audio volume control

Open-weight AI video model with synchronized audio, 20-second clips, and up to 120fps

Redo one line, one shot, one sound cue — without touching the rest of the clip

Multimodal video generation with native synchronized audio, text, image, or audio-driven input, and resolution up to 4K.

Speed-optimized LTX-2.3 variant: text-to-video and image-to-video with native synced audio, up to 4K and 20-second clips.

Camera moves that hold steady, not random AI drift

Animates static anime illustrations and manga panels into a steady 6-second, 720p character clip

Two images become a video's first and last frame; 768p runs 10s, 1080p runs 6s

High fidelity AI video generation from text or images

MiniMax Hailuo 2.3 Fast animates one photo into a 6-10 second video clip

AI talking avatar video from one photo and an audio clip

Fast, cheap AI video with synced dialogue

Animate one photo using the motion from an existing video.

Swap the character in any video — no reshoot needed

Re-sync a video's mouth movements to your own audio track for dubbing and redubbing

PixVerse V3.5 text-to-video with 20 built-in effect templates

Animate photos or prompts with camera motion and 5 styles

1080p text and image to video with camera movement control and start/end-frame guidance

PixVerse V5 cinematic text to video and image to video

Fast text to video and image to video generation for rapid iteration

More accurate multi-character lip-sync, cleaner motion in busy scenes, optional native audio, up to 1080p video

Localized video editing from a text prompt — swap products, backgrounds, or lighting without a reshoot, from $0.28/s

Single-frame image-to-video, 5 or 10-second clips, no Runway subscription required

Text-to-video and image-to-video in five aspect ratios, with 5, 8, or 10 second clips

1080p text- and image-to-video with a 1.2 to 12-second range, plus a 480p mode to preview cheaply before the final render.

Fast Seedance 1.0 Pro video generation for dance content

ByteDance's flagship video model — 30-second clips, 30 references, and prompt-based editing of an existing clip

One video model that generates, edits, and extends footage, with 8-frame anchoring and synced audio up to 1080p

Next generation AI video and audio model from OpenAI

OpenAI's 1080p Sora 2 tier — text-to-video or image-to-video, native audio, 4 to 20 second clips

Full-scene lip synchronization with global face understanding and obstruction handling

Google's original Veo: camera-directed video with first-and-last-frame anchoring

Cinematic video generation, now with native audio

Google's faster, cheaper Veo 3 tier — with start-and-end frame anchoring flagship Veo 3 lacks

Veo 3.1 cinematic AI video with native audio

1080p AI video with movement-amplitude control and built-in music

1080p AI video at 24fps, from a text prompt or first-and-last-frame images

AI video guided by 7 reference images

Faster Q2-tier video with adjustable motion, built-in music, and 7 reference images

Multimodal video generation with native audio and intelligent shot planning

Vidu's fast, low-latency video model with flexible sizing and optional audio

MoE video generation from text or images at 480p to 720p

Alibaba's original Wan video model — 5-10s clips from text or an image, with native audio

Multimodal video generation with multi-shot and native sound

The fast, distilled Wan2.6 Flash — lip-synced or silent image-to-video, up to 15 seconds

Alibaba's higher-end multimodal video model — keyframes, references, and stronger character consistency

OpenAI's image model for text-to-image and editing — three sizes, four quality tiers, up to 16 reference images.

OpenAI's newest image model — up to 16 reference images and native 4K output

Describe an edit in plain English — no masks, no sliders — then export up to 2K across nine aspect ratios

2K image generation from Kling AI, tuned for realistic textures and iterative image-to-image edits

High-resolution text-to-image and editing with multi-reference consistency

Google's Gemini 3.1 Flash Image — web-grounded search, 14 reference images, up to 2K output

Reasoning-first Seedream 5.0 tier, pairing built-in reasoning with live web search for accurate, current images

ByteDance's flagship image model: precise masked edits, 1K or 2K output, and up to 10 reference images in one request

The getvivix signature model — instant images in any style

Alibaba's unified image generator and editor — avatar customization, 9 reference images, and text in 12 languages

Alibaba pairs an optional thinking pass with 9-image reference grounding

Text to image model trained on licensed stock from Getty, Alamy, and Envato

Structured JSON prompts render reproducible images from licensed data

Edit photos from a text prompt — mask inpainting, outpainting, and generative fill

Preset-driven photo editing — 7 structured operations from relighting to outpainting, no prompt writing needed

DALL·E 2 online — no OpenAI account needed

DALL·E 3 high fidelity text to image generation API

Vibrant, high-contrast illustrative style with bold color palettes

Bright, energetic photographic style with vivid lighting

Heavy-shadow noir comic art, built for horror, thriller, and dystopian scenes

Dreamy photographic style with surreal, distant atmosphere

Warm, organic illustrative style with muted earth tones

Clean, editorial-style line illustrations with refined detail

High-contrast photographic style with dramatic light and shadow

Analog film photography style with natural grain and warm tones

Balanced, harmonious graphic illustrations with cohesive composition

Comic book and graphic novel style with strong ink lines and dramatic shading

Graphite-shaded creature and character art, built for fur, scales, and muscle detail

Cinematic golden-hour travel tones for your reference photo, 11 ratios up to 2K

Monochromatic illustrative style with warm café-inspired tones

Soft, desaturated illustration style built for modern branding and app visuals

Expressive, character-driven line art built for playful storybook and branding scenes

Soft, warm-lit photographic style with inviting golden tones

Swap a garment onto any person photo in 4 fast steps, keeping the face, pose, logos, and stitching intact

12B open-weight model, second only to FLUX.1 [pro] in output quality

Apache 2.0 open source FLUX.1 model with 1-4 step image generation

Open image editing model for fast iterative workflows

Precision image editing, not text-to-image — up to 2 reference images per edit

Unifies text-to-image generation and precise editing in one model — 2 reference images, 9 aspect ratios up to 21:9

Black Forest Labs' fine-tune of FLUX.1 dev, made with Krea AI to trade the oversaturated AI look for real photorealism

Black Forest Labs' well-rounded FLUX model for generating images and editing from a reference, in ten built-in aspect ratios.

Up to 4-megapixel FLUX images across nine aspect ratios, generated in about 10 seconds

The full 32B FLUX.2 model — precise control over prompts, references, and edits

The FLUX.2 tier that exposes manual sampling controls and up to 10 reference images for precise edits

Apache 2.0 open license — the fastest Klein model for real-time editing

Full 28-step image generation and editing from Black Forest Labs' undistilled klein foundation checkpoint

Undistilled foundation model for high-quality image generation and editing

KV-cache accelerated image generation and editing for real-time multi-reference workflows

Black Forest Labs' FLUX.2 tier that grounds images in live web search, with 8-image reference editing

High control FLUX.2 Pro image generation and editing

GPT Image 1 high fidelity image generation for GPT-4o

xAI's Grok Imagine model for text-to-image and image-to-image, the standard tier in its family

xAI's quality-focused image generation and editing — sharper realism, better text rendering, tighter prompt following

HiDream's balanced 28-step tier — quicker than the 50-step MIT-licensed Full, more steps than Fast

HiDream-I1 Fast for low latency text to image generation

MIT-licensed, 50-step flagship — the un-distilled source that Dev and Fast are distilled from

Sharp typography from Ideogram's per-style rendering model, released August 2024

Ideogram 3.0 text to image model for sharp design visuals

Ideogram's first open-weight model — precise typography, structured layout control, and transparent 2K output

Google's photorealistic Imagen 3 model for detailed, high fidelity images

Google's speed-tuned Imagen 3 variant, built for low-latency image generation

High speed Imagen 4 Fast text to image generation

High fidelity 2K text to image generation by Google

Google's most prompt-accurate Imagen model, for photorealistic detail and sharp text

Professional AI image generation with native 4K and refined visual control

Reasoning-based text to image generation with vibrant true-to-life color

A fast, low-cost Flux checkpoint for quick drafts and high-volume runs.

Photorealistic Flux checkpoint, hosted and ready — no download, no GPU, no setup.

Open-source text-to-image generation and image editing, built for fast, prompt-accurate 1K output.

Krea 2's biggest, rawest tier — over 2x Medium's size for stronger photorealism, film grain, and texture

Krea's smaller, faster tier — under half the size of Large, tuned for stable illustration

Gemini 2.5 Flash Image — the original Nano Banana, built for consistent-likeness photo edits

Lighter Nano Banana 2 image model for faster generation and editing workflows

Pruna AI's real-time text-to-image model — sub-second generation across nine size presets, square to 21:9 ultrawide

Edit with up to 4 reference images — composition and style stay locked

Apache 2.0 open source foundation model built for precise Chinese and English text in images

Alibaba's lighter next-gen model for clean in-image text and layouts

Edit an image with a text instruction, including bilingual Chinese-English text swaps that keep the font

Professional text-to-image model for brand and marketing design

Advanced design-focused image generation with enhanced control and fidelity

Native 4K x 4K, 16-megapixel image generation with layout-aware, region-by-region editing

One model for text-to-image generation and natural-language editing, up to 4K with 14-image reference fusion

From Stability AI, live here now — legible in-image text, multi-subject scenes, free to try

Wan2.5's still-image model — one generated frame, not a video clip

Edit photos with up to four reference images, or generate from a prompt, on Alibaba's Wan2.6 stack

Apache 2.0 open-source model from Alibaba — the full 6B Z-Image, built for creative control, not just speed

Fast photorealistic image generator with text control

Open-source (MIT) music generation with editable lyrics, song covers, and 50+ languages

Turns a lyric sheet into a full track faster — fewer denoising steps, same engine as the Base model.

English-only TTS at about 75ms latency, built for live streams, games, and voice apps

ElevenLabs' fast, affordable TTS model — 75ms latency, 32 languages, built for voice agents and bulk narration

The original Eleven TTS, English-only

Legacy multilingual TTS across 9 languages

ElevenLabs' expressive, 29-language voice model — built for emotional nuance over raw speed, no download required

Describe a song and get a finished track back — vocals or instrumental, structured section by section

First-generation low-latency Eleven TTS model, English only

First-generation low-latency Eleven TTS model, 32 languages

Dramatic multi-speaker dialogue TTS, 70+ languages

Expressive text-to-speech with audio tags, multi-speaker dialogue, and 70+ languages

Expressive, broadcast-ready text-to-speech: 73 voices, 15 languages, sub-250ms response time

Sub-130ms text-to-speech powering getvivix's avatars, agent, and chat voices

332 voices across 24 languages, with live emotion control and up to 50,000 characters per script

Clones a voice from a 3-second sample, then generates natural multilingual speech in 10+ languages at 97ms latency

Nine named preset voices with natural-language style control

Text-to-speech that designs a brand-new voice from your written description, no sample audio required

Text-to-speech with six voices, inline speech tags for pauses and emphasis, and 20 supported languages

Anthropic's top Claude model — Fable 5 sits above Opus in capability, with 1M-token context, vision, and adaptive thinking

Anthropic's fastest, most cost-efficient Claude tier — 200K context, 64K output tokens, built for sub-agent dispatch

Free to start, 100+ other AI models on one account — Anthropic's top Claude 4.7, 1M-token context, xhigh thinking tier

Anthropic's daily-driver Claude — reads images, six adjustable thinking levels, and 65,536-token replies

Open-weight MoE model, MIT-licensed, with a 1M-token context and switchable reasoning depth

Multimodal understanding model for text, image, video, and audio with a 1M-token context

Google’s fast, low-cost Gemini 3.1 tier for high-volume translation, transcription, and document extraction

Google's multimodal reasoning model with a 1M-token context window

Z.ai's open-weight value LLM — 200K context, 73.8% on SWE-bench Verified

Z.ai's open-weight flagship LLM — 200K context, MIT license, 8-hour autonomous agent runs

OpenAI's flagship model — 1M-token context, native computer-use, and 33% fewer factual errors than GPT-5.2

OpenAI's compact GPT-5.4 sibling: 400K context, vision input, and computer use for coding agents and subagent workflows

Ultra-low-latency LLM for high-volume classification, extraction, and lightweight automation

OpenAI's newest flagship LLM — deepest reasoning, computer-use, 1M+ context

262K context, configurable reasoning, and an 80.2 SWE-Bench Verified score

Turns an image into a caption, guided by an optional prompt from a Mistral 7B backbone

MiniMax's original agentic-coding and office-task model — released before the longer-context M2.7 and faster M2.7 Highspeed

MiniMax's 230B-parameter agent model also drives an AI Character's dialogue on getvivix

Faster throughput for agentic coding and tool‑driven automation

Guess your age with AI

A vision-language model that reads an image and writes a caption for it

Caption or answer questions about any photo instantly — no download, GPU, or license required

Open-weight vision-language model that reads and describes images, charts, and documents

Vision Transformer model that classifies facial age into brackets with a confidence score

2x or 4x image upscaling that preserves transparency, for illustrations, product shots, and UI assets

Remove a video's background in real time — no green screen needed

2x or 4x video upscaling that keeps transparency and fine detail intact

Gemini 3 Pro's image model — 2K generation with 14-image blending and built-in scene reasoning

Upscales AI-generated images up to 8 megapixels with detail and realism enhancement controls

High precision multi image editing for daily workflows

The lowest-cost Riverflow 1.1 tier, with the same reference-image range and aspect ratios as Pro

Layout-safe image editing with 10 reference photos and 11 aspect ratios

Fast lightweight Riverflow 2 for product accurate images

Generates, edits, and upscales product renders up to 2K resolution

Balanced Riverflow 2 image model for pro workflows

Fast production image generation with reference-based super resolution and font control

ByteDance's 2K-4K image model with 14-image reference fusion for consistent characters and sharp small text

Turn a photo or text prompt into a textured 3D mesh in your browser — no API key, no install

Photo or prompt to a 3D mesh in seconds — quick iterations, no API key needed

Image- or text-to-3D — PBR-textured, production-grade meshes for games and AR/VR

Turn a single photo into a textured, physically accurate 3D mesh with sharp geometric detail

Photo or prompt to a clean 3D mesh