Répertoire d'outils IA

A text-to-image AI known for its painterly, stylized aesthetic and cinematic lighting. Accessible through Discord and a dedicated web interface, Midjourney has become a signature tool for concept artists, illustrators, and art directors seeking high-visual-impact imagery with minimal prompt engineering.

OpenAI's third-generation text-to-image model with industry-leading prompt adherence, readable text rendering, and deep integration into ChatGPT. DALL-E 3 is designed to follow long, specific instructions faithfully, making it a strong choice for narrative illustration and editorial work.

The open-source foundation of the modern AI art ecosystem, developed by Stability AI. Stable Diffusion runs locally on consumer GPUs, spawning thousands of community-trained models, LoRAs, and interfaces such as Automatic1111, ComfyUI, and Fooocus. Its flexibility makes it the professional's choice.

Adobe's generative AI family integrated into Photoshop, Illustrator, Express, and the standalone Firefly web app. Trained exclusively on Adobe Stock, licensed content, and public domain material, Firefly is positioned as the "commercially safe" AI art tool with indemnification for enterprise customers.

A generation platform originally focused on game asset creation that has grown into a full-featured art studio. Leonardo offers fine-tuned models, real-time canvas editing, 3D texture generation, and image-to-video features, with a strong free tier that makes it accessible for hobbyists.

The flagship open-weight model family from Black Forest Labs, founded by the original Stable Diffusion team. Flux models (Schnell, Dev, Pro) set a new bar for prompt adherence, photorealism, and readable text in open image generation, and have been adopted across the open-source ecosystem.

A text-to-image platform specialized in accurate typography, logos, and poster-style designs. Ideogram's standout capability is rendering legible, well-placed text inside images, making it a favorite for designers working on flyers, social graphics, and brand-forward compositions.

A real-time AI canvas that generates images as you sketch, type, or move shapes. Krea blends Stable Diffusion, Flux, and custom models with live feedback, making ideation feel like drawing with a responsive collaborator. It also offers upscaling, video, and 3D tools.

An AI design platform focused on vector graphics, brand consistency, and professional design workflows. Recraft generates SVG-ready vector illustrations, infographics, icons, and mockups, and its style-reference feature locks brand aesthetics across large batches of outputs.

Microsoft's free DALL-E 3-powered image generator, integrated into Bing Search and Copilot. It offers the quality of DALL-E 3 at no cost, making it an ideal entry point for curious beginners who want to experiment with state-of-the-art generation without subscriptions.

The most widely used AI music generator, capable of producing fully mixed songs with vocals, lyrics, and instrumentation from a text prompt. Suno's recent models (v3.5, v4) approach commercial production quality and support extended song lengths, custom lyrics, and stem downloads.

A music generation platform from ex-Google DeepMind researchers that emphasizes sonic quality, vocal realism, and extensibility. Udio offers fine-grained remixing, inpainting of specific song sections, and a growing feature set aimed at professional music producers.

An AI composer designed specifically for instrumental, orchestral, and soundtrack music. AIVA outputs editable MIDI and sheet music, making it a powerful starting point for film composers, game music producers, and classical composers who want AI to draft arrangements.

A royalty-free AI music platform targeted at content creators, with simple genre/mood selectors and customizable stems. Soundraw generates unlimited tracks under a subscription and lets users edit song structure, instruments, and energy levels with visual controls.

A consumer-friendly platform that turns anyone into a music artist by generating full songs in seconds and offering one-click distribution to Spotify, Apple Music, and other streaming services. Boomy has helped release millions of user tracks into commercial streaming.

An AI music engine that generates continuous, royalty-free soundtracks using curated loops and algorithmic arrangement. Mubert is built for streaming contexts (Twitch, games, apps) and offers both a consumer-facing app and a developer API.

An AI music generation platform combining prompt-based creation with a curated library of stems. Loudly targets content creators and offers a browser-based studio where users can generate, remix, and export tracks in common formats.

One of the earliest AI music composition platforms, now integrated into Shutterstock's stock media ecosystem. Amper lets users generate custom tracks with mood, genre, and length controls, with straightforward licensing for use in commercial media.

The flagship AI video platform, Runway's Gen-3 and Gen-4 models power professional filmmakers, agencies, and studios. Beyond text-to-video, Runway offers image-to-video, motion brush, camera controls, green-screen, and a full suite of editing AI, making it the most complete creative video AI.

A playful, fast-moving AI video platform known for fun effects like "Pikaffects" (explode, melt, squish) and strong image-to-video animation. Pika is approachable for beginners while offering lip-sync, extensions, and camera controls for more serious work.

Luma Labs' video model, praised for realistic physics, smooth motion, and strong coherence over longer clips. Dream Machine offers text-to-video, image-to-video, keyframe-based control, and extend features, and is one of the fastest-improving models in the space.

Kuaishou's video generation model, Kling has emerged as a serious competitor to Runway and Sora, with notable strengths in human motion, longer clip lengths (up to 2 minutes), and affordability. Available via the Kling web app and API.

OpenAI's flagship video model, capable of generating highly realistic, narratively coherent clips up to 20 seconds at 1080p. Sora is integrated into ChatGPT Plus and Pro tiers and offers a dedicated storyboard-style editor with scene extension and remixing.

The leader in AI avatar video, Synthesia turns written scripts into professional presenter videos featuring realistic AI avatars in 140+ languages. It's widely used for corporate training, learning content, and marketing at enterprise scale.

A direct competitor to Synthesia with strong AI avatar quality, real-time translation, lip-sync in 175+ languages, and the ability to create custom avatars from a short video. HeyGen is popular with creators, marketers, and educators.

A video AI platform focused on "talking head" animation from a single photo, used heavily for interactive avatars, virtual agents, and lightweight presenter content. D-ID offers a web studio and a robust API for embedding avatars in apps and websites.

Anthropic's AI assistant, Claude is widely regarded as the strongest model for long-form writing, nuanced creative work, and thoughtful collaboration. Claude 4.7 and the 1M context window make it especially powerful for editing books, analyzing transcripts, and sustained creative projects.

The most widely known AI assistant, ChatGPT combines GPT-4o and GPT-5 models with DALL-E 3 image generation, Sora video, Advanced Voice, code interpreter, and web browsing. It's the Swiss Army knife of AI assistants and the default entry point for most creators.

Google's flagship AI assistant, Gemini integrates deeply with Google Docs, Gmail, Search, and Workspace. Gemini 2.5 Pro and Ultra models offer massive context windows, strong multimodal capabilities, and live web access, making it a productivity powerhouse.

Google's AI research notebook, NotebookLM grounds every response in sources you upload (PDFs, Google Docs, websites, videos). Its standout "Audio Overview" feature generates podcast-style conversations from your sources, making it a favorite for research and study.

A writing AI designed specifically for fiction authors, with features like Story Engine, Canvas, character development, plot brainstorming, and genre-aware rewrites. Sudowrite integrates multiple LLMs under the hood and is shaped by working novelists.

An enterprise AI writing platform focused on marketing content, brand voice, and team collaboration. Jasper offers brand-voice templates, campaign workflows, plagiarism checking, and integrations with Surfer SEO, Zapier, and major marketing stacks.

A go-to-market AI platform combining writing templates, workflow automation, and sales/marketing agents. Copy.ai started as a copywriting tool and has evolved into a full workflow platform for revenue teams, though it still offers strong per-task copy generation.

Canva's Magic Studio brings AI to its popular design platform with Magic Design (instant templates), Magic Write (copy), Magic Media (image/video generation), background removal, and brand-aware generation. It's the most accessible design AI for non-designers.

Figma's growing suite of AI features for product and UI designers, including first-draft generation, layer renaming, auto-layout suggestions, prototype generation, and visual search. Figma AI is designed to accelerate existing design workflows rather than replace them.

An AI UI generator that turns text prompts into editable Figma designs. Galileo AI is useful for rapid concepting of mobile screens, web pages, and product flows, and bridges the gap between an idea and a polished design canvas.

A rapid UI design tool that turns sketches, screenshots, and text prompts into editable mockups and interactive prototypes. Uizard is aimed at non-designers, PMs, and founders who need to visualize ideas quickly without mastering Figma.

Framer is a no-code website builder with deep AI integration for generating full websites from prompts, localizing content, writing copy, and translating designs. Framer AI is well-suited to designers who want to ship polished marketing sites fast.

A leading text-to-3D and image-to-3D platform producing game-ready meshes with PBR textures. Meshy is used by indie game devs, AR/VR creators, and 3D artists for rapid asset creation, with support for common 3D formats (OBJ, FBX, GLB, USDZ).

Luma Labs' 3D capture and generation platform. Luma captures real-world scenes as NeRFs or Gaussian splats from phone video, and "Genie" generates 3D models from text prompts. Widely used in film previs, VFX, and immersive content.

An image-to-3D platform for game and product developers that combines AI with human QA to produce production-grade meshes. Kaedim targets studios with strict quality requirements and integrates directly with Unity, Unreal, Blender, and Maya.

Tripo AI generates 3D models from text or images in seconds, with a free web interface and competitive quality. It's a popular choice for hobbyists, game jammers, and 3D printers who want quick, no-friction asset creation.

Photoshop's deep integration of Firefly powers Generative Fill, Generative Expand, Remove Tool, and Neural Filters. These features transform photo editing workflows, making tasks like background removal, object removal, and image extension a single click.

Topaz's photo and video AI suite (Photo AI, Gigapixel, Video AI) specializes in upscaling, denoising, sharpening, and restoration. Preferred by professionals for archival work, landscape and wildlife photography, and salvaging low-quality footage.

A colorization AI specialized in bringing black-and-white photographs to life with context-aware, photorealistic color. Palette.fm is used by archivists, historians, families, and publishers to restore and reinterpret historical imagery.

A mobile photo editor by Prisma Labs known for its "Magic Avatars" feature that generates stylized AI portraits from a set of selfies. Lensa also offers background removal, skin retouch, and other common mobile photo AI features.

Skylum's AI-first photo editor positioned as a Lightroom/Photoshop alternative for creators. Luminar Neo offers Sky AI, Relight AI, Portrait Bokeh AI, Enhance AI, and more, wrapped in a modern, approachable interface with one-time or subscription pricing.

Google's Gemini 2.5 family is natively multimodal, processing text, images, audio, video, and code in a single context. For creators, this means uploading a reference image, a voice memo, and a brief, then getting coherent cross-media analysis or creative output.

OpenAI's omni models (GPT-4o and successors) handle text, image, and voice natively in real time. The Advanced Voice Mode, vision input, and integrated DALL-E and Sora generation make it a creative hub for multimodal ideation and production.

Claude 4.x models handle text and images natively, with a 1M-token context window (on Opus 4.7) that can ingest entire books, codebases, and visual archives. Claude's strengths in nuanced writing and careful reasoning extend to multimodal analysis and critique.

Google DeepMind's third-generation video model, generating 1080p clips up to 8 seconds with native synchronized audio (dialogue, ambient sound, foley). Veo 3 leads the field on physics realism, multi-shot consistency, and lip-sync. Available through Vertex AI and the Gemini app for Pro/Ultra subscribers.

Black Forest Labs' professional image model, succeeding the original FLUX with significantly improved prompt adherence, photorealism at high resolution, and sub-10-second generation on managed APIs. The Ultra tier supports 4MP outputs and is widely considered the strongest open-weights image model in production.

A video-generation platform focused on cinematic camera control — preset moves like dolly-in, orbit, crane, and "Bullet Time" — applied to user-uploaded reference images. Built around the insight that filmmakers want directable camera language more than longer durations.

MiniMax's video-generation platform from China, notable for natural human motion, expressive faces, and competitive pricing. Often surfaces as a strong alternative when Runway and Kling produce stiff or unrealistic character motion.

A high-fidelity image model from a small independent team, known for exceptional typography rendering and design-quality output. Often produces magazine-cover-grade images on first try, with text accurately rendered in-image — a long-standing weak spot for most generators.

Alibaba's open-source video-generation model, released with full weights for self-hosting. Wan 2.2 is the strongest open video model available and is used by smaller studios who want to run video generation locally without paying per-second API fees.