Fast drafts, fast finals
Draft at 1K in seconds, then ship the winner at 2K or 4K. Ideas stop waiting on renders.
Hot
MiniMax H3 Max AI Video Generator
The reference-native tier of MiniMax H3 — feed stills, clips, and audio cues to keep identity and style locked across every generated shot.
AIモデル 5 クレジット
プロンプト0/2000
著作権で保護されたキャラクター・ロゴ・音楽や、敏感・露骨な内容は避けてください。ブロックされる場合があります。
解像度
アスペクト比
バッチ生成1
推定コスト：5クレジット
作品がここに表示されます
モデルを選び、ショットを描写して「生成」を押してください。
Why Nano Banana AI
One model. Every visual you described.
The Nano Banana 2.1 model reads prompts like an art director — subject, composition, light, and typography — and renders images that look finished, not synthesized.
Multi-reference editing
Feed up to eight reference images — subjects, styles, and scenes stay consistent while you change exactly what you asked.
Readable text rendering
Posters, packaging, and signage come out with crisp, legible type — one of the strongest text-in-image models available.
Beyond images
The same studio hosts frontier video models — Muse Video, Seedance 2.5, MiniMax H3 — for when your still needs to move.
How it works
From sentence to visual in three steps
Write what you want to see — subject, style, lighting, mood. Drop reference images if you have them.
00:0000:1500:30
Made for
Built for people who ship visuals
Creators & studios
Concept art, storyboards, thumbnails, and key visuals — without waiting on a render farm or a reshoot.
Marketers
Turn product shots and copy into scroll-stopping creatives for every placement and ratio.
E-commerce teams
Reshoot catalog photos in new scenes, swap backgrounds, and produce seasonal variants on demand.
Designers
Posters, mockups, and text-accurate layouts — iterate in natural language instead of rebuilding layers.
Capabilities
Every frontier model in one studio
Switch engines without switching tools — Nano Banana 2.1 plus the strongest public image and video models, side by side in the Nano Banana AI studio.
Multi-reference editingCharacter consistencyText rendering1K–4K outputConversational editsStyle transferVideo generationBatch runs
Muse VideoNewネイティブ音声付きフラッグシップ動画
Seedance 2.5Hotマルチ参照対応Seedance（ネイティブ音声）Seedance 2.0前世代Seedance（最大1080p）Seedance 2.0 FastFastドラフト向け最速Seedance
MiniMax H3最大2Kの高速テキスト/画像動画化MiniMax H3 MaxHot参照駆動の動画一貫性
Veo 3.1Googleのシネマティック・ドラフトエンジン
Wan 3.0柔軟な長さとアダプティブフレーミング
Kling 3.01つのプロンプトで劇的なモーション
GPT Image 2.5New4Kまでのシャープな文字と精密編集GPT Image 24K対応の万能画像生成Nano Banana 2.1New高速ドラフトと精密編集、最大4KNano Banana ProHot4K高忠実度画像出力Nano Banana 2高速ドラフトと会話型編集Nano Banana 2 LiteFast手早く低コストで下書きできる軽量版FAQ
Questions, answered
What does MiniMax H3 Max add over H3?
H3 Max is built around references: up to 9 reference images plus video and audio cues, keeping characters and art direction consistent between clips.
Which inputs work best?
Three to five clean reference stills covering different angles of the subject give the strongest consistency.
What resolution does H3 Max output?
768P by default, up to 2K for final renders.
Can H3 Max extend an existing clip?
Use a video reference to steer motion and composition — the model uses it as a guide rather than a continuation.