UGC ads, short series, AI influencers — rendered from $0.03. No subscription.

A 10-second vertical clip with speech and ambient sound on Gemini Omni 1.1 Flash costs $1.00 at 720p, and a 360p draft of the same clip is billed at $0.03 per second. You set the exact length and pay per second, so a short hook costs less than a full review. Grok Imagine Video 1.5 ($0.14 per second at 720p) and Wan 3.0 ($0.10 per second) also render speech, and silent Veo 3.1 clips start at $0.10. There is no subscription: you top up a balance and each render is charged from it.

Yes — the person in a BananaBanana clip can speak your script and hold your product. Quote the line in the prompt and the model renders the voice in the same pass as the picture. For your real product, attach reference photos or generate an opening frame with the product first and animate that frame, so the packaging keeps its shape.

It depends on the content, the jurisdiction and the platform. The FTC applies its endorsement and fake-review rules to AI-made testimonials, the EU AI Act sets transparency duties for deepfakes, and ad platforms have their own synthetic-media labels. BananaBanana does not add a visible watermark, so disclosure is your responsibility. Do not present a generated person as a real customer.

Wan 3.0 renders up to 30 seconds in one pass. Gemini Omni 1.1 extends a finished clip up to 40 seconds in total. Veo 3.1 adds 7 seconds per Extend, and chained extensions reach 148 seconds with visual continuity.

Generate a reference frame for each character once, then attach those frames as references to every scene. You can also start a clip from an approved first frame, so the scene opens exactly where you decided.

No — BananaBanana does not write, edit or publish a series; it renders the shots and the voice-over you describe, and your agent can assemble them locally with ffmpeg. Scripts, editing, publishing and the audience are yours, and nothing here is a promise of views or income. It is a production tool for people who already make short series.

On Nano Banana, attach up to 14 reference photos of the character per generation and describe the new scene. You can render up to 8 frames per prompt and keep the best ones. Video models accept references too, with their own limits per model, so stills and clips show the same face.

Yes — a brand can use an AI influencer or virtual model made on BananaBanana commercially, within the model providers' terms. The AI model must be an invented character: do not recreate a real person's face or voice without consent, and disclose synthetic media where the law or the platform requires it.

No — adult content is not allowed: AI influencer content on BananaBanana is SFW feed content. The Relaxed Filter gives more room for fashion, swimwear and art, but explicit content is blocked by the model providers and is not what the platform is for.

Yes — BananaBanana turns a product photo into a video. Upload the photo as the first frame, describe the camera move, and the clip starts from your exact image. Some models also accept a last frame, so you control where the shot ends. Clips start at $0.10.

Starting from your photo as the first frame is the most reliable way to keep geometry. Small text on labels can still drift in motion, so check the result and re-render if needed. Short, slow camera moves hold detail best.

Grok Imagine Video 1.5 follows a first frame very closely — a first frame works at 720p and 1080p, reference images at 720p, clips run 4–15 seconds and sound is included. Gemini Omni 1.1 Flash and Wan 3.0 also accept a last frame and go above 720p, Wan 3.0 gives takes up to 30 seconds, and Veo 3.1 is the top tier for native 4K. For the still itself the usual choice is Nano Banana Pro — or GPT Image 2.5 and Qwen Image 3.0 Pro when the packshot carries readable text.

Yes — footage can be generated automatically from n8n or your own agent. The BananaBanana remote MCP server exposes image, video and speech tools to any MCP client, including n8n, Claude Code, Cursor and Gemini CLI. Calls are charged from the same balance as the web app, and failed generations are refunded.

Images start at $0.03 each, and one prompt can return up to 8 frames. Vertical video clips start at $0.10. The total depends on the model, the settings, the number of outputs and retries.

Monetisation of automated videos is decided by YouTube, not by BananaBanana. YouTube does not monetise mass-produced, repetitive content. BananaBanana supplies the raw material: frames, B-roll, clips and voice-over. The editing and the originality of the channel are what the platforms judge.

BananaBanana is a pay-as-you-go AI image, video and speech generator with 14 models from Google, OpenAI, Alibaba and xAI. Images start at $0.03 and video clips at $0.10. Write a prompt, optionally attach reference photos or a first frame, see the price before you render, and get up to 8 images per prompt. Top up with crypto from $1 depending on coin and network, or by card, PayPal or SEPA from $20 — no subscription, and no card needed to sign up. Every new account gets a $0.20 welcome balance.

14 models: 6 for images, 7 for video and one for speech. Images — Nano Banana 2 Lite (flat $0.03 at 1K, the cheapest way to iterate), Nano Banana 2 (from $0.03, up to 4K), Nano Banana Pro (from $0.11, the most detailed portraits and product shots, up to 4K), GPT Image 2.5 Flare and Sunburst from OpenAI (from $0.05, the pick when the picture must contain readable text; Flare is the fast tier, Sunburst the precision tier), and Qwen Image 3.0 Pro from Alibaba (from $0.04, dense layouts and small text, 1K and 2K). Video — Veo 3.1 (from $0.70), Veo 3.1 Fast (from $0.35) and Veo 3.1 Lite (from $0.10): 4, 6 or 8-second clips, optional sound, up to 4K on Veo 3.1 and Fast; Gemini Omni 1.1 Flash (from $0.09): 3–10 seconds with always-on sound and speech, 360p drafts to 4K, editing and scene extension up to 40 seconds; Gemini Omni Flash 1.0 (from $0.40), the previous generation, kept for projects that started on it; Wan 3.0 from Alibaba (from $0.20): up to 30 seconds in a single pass, first and last frame, sound you can switch off; Grok Imagine Video 1.5 from xAI (from $0.56): 4–15-second clips with native sound, five aspect ratios. Speech — Gemini 3.1 Flash TTS from Google ($0.01 per started 200 characters): 30 voices, one or two speakers, in the web studio and through MCP.

Every new account gets a $0.20 welcome balance with no card required — enough for several images or a short draft clip. After that you pay per render. Images: Nano Banana 2 Lite is a flat $0.03; Nano Banana 2 runs from $0.03 to $0.13 (512px to 4K); Nano Banana Pro from $0.11 to $0.20; GPT Image 2.5 Flare and Sunburst from $0.05 to $0.18 (1K to 4K); Qwen Image 3.0 Pro $0.04 at 1K and $0.08 at 2K. Video: Veo 3.1 Lite from $0.10, Veo 3.1 Fast from $0.35, Veo 3.1 from $0.70. Gemini Omni 1.1 Flash is billed per second by resolution — $0.03 at 360p, $0.10 at 720p — so a 10-second 720p clip with sound is $1.00. Wan 3.0 is $0.05–$0.20 per second by resolution, Grok Imagine Video 1.5 is $0.14 per second at 720p and $0.25 at 1080p. The full tables are in the Pricing section above. Top up $50+ in crypto for a 5% bonus or $100+ for 10%; card, PayPal and SEPA payments are added to your balance without a bonus. Speech on Gemini 3.1 Flash TTS is $0.01 per started 200 characters. Unused balance never expires, and failed renders are refunded.

BananaBanana accepts both crypto and card payments. Crypto: USDT (ERC-20, BEP-20, SOL, TON), USDC (SOL, BASE, ERC-20, BEP-20), DAI, Bitcoin (BTC), Ethereum (ETH), Solana (SOL), BNB, Litecoin (LTC), Tron (TRX), TON, XRP, Dogecoin (DOGE), and others — click "Top Up", choose an amount (minimum from $1 depending on coin and network), select your cryptocurrency and network, and receive a permanent wallet address that works for all future deposits; the balance is credited automatically after blockchain confirmation. Card: Visa and Mastercard, a PayPal balance or a SEPA transfer in the EU, from $20 to $2,000 per payment, handled inside the PayPal window so card details never reach our site, and credited as soon as the payment clears. Whichever way you pay, it is a one-off top-up: no subscription, no saved card, no auto-renewal — the balance sits there until you spend it. Crypto is the route we built first and still the lighter one: from $1 instead of $20, no bank in the loop, and fees of a few cents on networks like TON or Solana. Volume bonuses apply to crypto top-ups only: +5% on $50+, +10% on $100+; card, PayPal and SEPA payments are credited at face value.

BananaBanana is available worldwide with no geographic restrictions. Crypto top-ups work from any country without a local bank account, and card, PayPal and SEPA payments run through PayPal wherever it operates. The interface is available in English, Arabic, Spanish, French, Hindi, Indonesian, Japanese, Portuguese, Russian, Thai, Turkish, Ukrainian, Urdu, Vietnamese, and Chinese. All AI models and features are accessible to every user regardless of location.

Images take 1–2 minutes; videos take 2–5 minutes depending on model and settings. Veo 3.1 Fast and Veo 3.1 Lite are generally quicker. You can close the page during generation and come back later — results will be waiting in your account.

BananaBanana includes built-in character consistency — upload a portrait photo as a character reference, describe your scene, and the AI preserves facial features while generating the person in a new context, outfit, or art style. Attach up to 14 reference photos (characters and objects) per generation for consistent character creation across multiple images. Ideal for AI influencer content, social media visuals, marketing materials, or any project requiring the same character across different scenes.

Yes — BananaBanana is built for AI influencer and content creator workflows with Reference mode, character consistency, and up to 8 images per prompt. Create a consistent AI character and generate that same character in different scenes, outfits, and locations — ideal for Instagram, TikTok, and other platforms. Nano Banana Pro produces highly realistic, professional-quality portraits and lifestyle photos. Unlike the web version of Google's Gemini, BananaBanana generates content without any visible watermark. Note: images and video from Google models (Nano Banana, Veo, Gemini Omni) carry Google's invisible SynthID watermark — it does not affect quality but may be detected by some platforms' AI content identification systems. Enable Social Media Optimization to prepare your images for clean publishing on social platforms without quality loss. Pay-as-you-go pricing, topped up with crypto or a card, makes it accessible for solo creators and agencies.

Yes — generate up to 8 images per prompt to get multiple variations at once, ideal for e-commerce product visuals, print-on-demand designs, social media content, and marketing campaigns. Reference mode ensures visual consistency when you need the same character, product, or style across different scenes.

Images: Nano Banana 2 and Pro render 512px to 4K (Pro starts at 1K) in ten aspect ratios from 1:1 to 21:9; Nano Banana 2 Lite is 1K only; GPT Image 2.5 offers 1K, 2K and 4K; Qwen Image 3.0 Pro 1K and 2K. Up to 8 images per prompt. Video: Veo 3.1 and Veo 3.1 Fast — 4, 6 or 8 seconds at 720p, 1080p or 4K; Veo 3.1 Lite — the same durations up to 1080p. Gemini Omni 1.1 Flash — any length from 3 to 10 seconds in one-second steps, 360p, 720p, 1080p or 4K, extendable to 40 seconds. Gemini Omni Flash 1.0 — 4, 6, 8 or 10 seconds at native 720p, or 6, 8 or 10 seconds as an upscaled 1080p. Wan 3.0 — 4, 6, 8, 10, 15, 20 or 30 seconds at 480p, 720p or 1080p. Grok Imagine Video 1.5 — 4 to 15 seconds at 720p or 1080p. Every video model renders 16:9 and 9:16; Grok also does 1:1, 3:2 and 2:3.

Yes — BananaBanana supports multi-turn AI image editing. Select a generated image and describe changes in natural language; the AI modifies the image while preserving the original composition. Chain multiple edits in a conversation-style workflow to refine results step by step — similar to chatting with an AI assistant, but for image editing.

Social Media Optimization is a one-click option that processes your generated images for seamless publishing on social platforms — reducing the chance of automatic AI content labels without quality loss. Just toggle it on in the generator settings before generating. Works with any resolution and model. Google's invisible SynthID watermark is preserved, keeping your content compliant with AI transparency standards.

Pick the video model by what the clip has to do: Veo 3.1 for polished final renders, Gemini Omni 1.1 Flash for talking clips billed per second, Wan 3.0 for long single takes, Grok Imagine Video 1.5 for clips that follow a first frame closely. Veo 3.1 (from $0.70) is the top Veo tier for final renders, Veo 3.1 Fast (from $0.35) is the practical default, and Veo 3.1 Lite (from $0.10) is the cheapest Veo tier for testing an idea; all three render exact 4, 6 or 8-second clips, sound is optional, and Veo 3.1 and Fast go up to native 4K and chain with Extend. Gemini Omni 1.1 Flash ($0.10 per second at 720p, $0.03 at 360p) is the choice for talking clips: sound is always on, quoted lines in the prompt are spoken by the character, you pay for the exact 3–10 seconds you need, it accepts a first frame, a last frame and reference images, edits a finished clip from a text instruction, and extends a scene to 40 seconds. Gemini Omni Flash 1.0 is the previous generation at native 720p and upscaled 1080p, kept for continuity. Wan 3.0 ($0.10 per second at 720p) is the only model that renders up to 30 seconds in a single pass, with first and last frame control, seeds and sound you can switch off. Grok Imagine Video 1.5 ($0.14 per second at 720p, $0.25 at 1080p) renders 4–15-second clips with native sound, follows a first frame very closely, and supports square and 3:2 formats.

Yes — every video model here can render sound in the same pass as the picture. On Gemini Omni 1.1 Flash and Omni Flash 1.0 audio is always on and included in the per-second price; quote a line in the prompt and the character speaks it (check the take for accuracy), which makes Omni the usual choice for talking-head and UGC clips. The three Veo 3.1 models generate ambient sound, music, effects and dialogue when you switch audio on, with an optional audio prompt to guide the sound design; silent Veo clips cost less. Wan 3.0 includes sound at the same price, and you can switch it off for free. Grok Imagine Video 1.5 always renders a soundtrack and voices spoken lines. For a separate voice-over, Gemini 3.1 Flash TTS offers 30 voices and one or two speakers, in the web studio and through MCP.

Yes — every video model on the platform accepts a first frame: upload a photo, describe the motion, and the clip opens on your exact image. Veo 3.1, Gemini Omni 1.1 Flash and Wan 3.0 also accept a last frame, so you control where the shot ends or build a seamless loop. Grok Imagine Video 1.5 follows the first frame especially closely (image inputs there work at 720p). Instead of a fixed first frame, most video models accept reference images of a character or product and stage the scene themselves; the limits differ per model and are shown in the generator. Veo 3.1 and Veo 3.1 Fast go up to 4K, Omni 1.1 up to 4K, Wan 3.0 and Grok up to 1080p.

Yes — BananaBanana extends finished AI videos in three ways. Veo 3.1 and Veo 3.1 Fast: click Extend on a finished clip and continue it with a new prompt; the model keeps visual continuity, and chained extensions reach 148 seconds. Gemini Omni 1.1 Flash: append 3–10 seconds at a time to a finished clip, up to 40 seconds in total (each extension is priced on the full resulting length, shown before you render), or edit the clip with a text instruction while keeping the scene and characters; you can also send the first 10 seconds of any finished video — a Veo clip or your own upload — for a video-to-video restyle at 720p. Wan 3.0 rarely needs extension, because a single pass already runs up to 30 seconds; over MCP it can also read a source clip and shoot the next scene with the same characters and setting — a story continuation, not a frame-exact one. Grok Imagine Video 1.5 renders standalone clips of 4–15 seconds.

In a single pass: 30 seconds on Wan 3.0, 15 seconds on Grok Imagine Video 1.5, 10 seconds on Gemini Omni, 8 seconds on Veo 3.1. With extension: up to 40 seconds on Gemini Omni 1.1 Flash, adding 3–10 seconds at a time, and up to 148 seconds on Veo 3.1 and Veo 3.1 Fast by chaining Extend. Each pass or extension is priced separately and shown before you render.

The main difference is the billing model and the range. Many generators sell a monthly subscription with credits that reset; here there is no subscription at all — images from $0.03, video from $0.10, charged per render from a balance that never expires. Instead of one vendor's model you get 14 from four: Google's Nano Banana, Veo 3.1 and Gemini Omni, OpenAI's GPT Image 2.5, Alibaba's Wan 3.0 and Qwen Image 3.0 Pro, and xAI's Grok Imagine Video 1.5. That covers native 4K video, clips with speech, single takes up to 30 seconds, scene extension up to 148 seconds, and images with readable text. There are no visible watermarks, fair-use rate limits replace monthly quotas, and the same balance works from the web app and from AI agents over MCP.

BananaBanana provides a more convenient and affordable way to access Google's Gemini image models and Veo 3.1 video models. Key advantages over Gemini: no visible watermarks on generated content, Relaxed Filter option for fewer content blocks, up to 8 images per prompt, 4K video with audio generation, video extension that chains clips up to 148 seconds, character consistency via reference photos, and crypto or card payments with no Google account required. Prices start at $0.03 per image — comparable to or cheaper than API pricing, without needing to set up billing or manage API keys.

BananaBanana offers negative prompts (up to 1,000 characters) to exclude unwanted elements, seed values for reproducible results, Social Media Optimization (SMO) for seamless social media publishing, up to 4 video variants per prompt, and image output in PNG, JPEG, or WebP formats. A fixed seed keeps results close between runs of the same prompt — useful for consistency testing and A/B comparisons — but models do not guarantee a pixel-identical repeat. These professional controls give creators and developers precise command over AI-generated content.

BananaBanana offers a Relaxed Filter toggle in the generator settings that provides significantly more creative freedom than Midjourney, GPT Image, Grok Imagine, or the web version of Gemini. When enabled, BananaBanana sends Gemini's safetySettings API parameter with the OFF threshold — this does not turn moderation off entirely, it only disables the prompt-level safety filter exposed through the API. Gemini also runs a second, internal filter whose behavior cannot be controlled via the API, so extreme content is still blocked. Even so, the toggle gives noticeably more room for legitimate creative work like art, fashion photography, or mature storytelling — without asking you to break any rules. The toggle can be switched on or off for each generation.

Yes — generated images and videos are yours to use for personal and commercial purposes, including ads, social media, product visuals, AI influencer accounts, e-commerce listings and print-on-demand. Content is subject to the usage policies of the model provider that rendered it (Google, OpenAI, Alibaba or xAI), and disclosure of synthetic media where the law or a platform requires it is your responsibility.

Creating an account takes only an email address — no name, phone number or postal address, and no card details on our side: crypto top-ups go to a wallet address, while card, PayPal and SEPA payments happen inside PayPal's own window, so card numbers never reach us; we keep only the transaction records. Like any hosted service we also log technical data: IP address, browser and usage history; the Privacy Policy lists all of it. Prompts and reference images are sent to the provider of the model you chose — Google, OpenAI, Alibaba or xAI, directly or through an API gateway — with no account identifier attached, and we never sell them or use them to train models. Generated images and videos stay in your private account for 30 days and are then deleted automatically; you can also delete them manually at any time.

Yes — BananaBanana offers a three-tier affiliate program with uncapped earnings. Create up to 3 personal promo codes — when someone uses your code, they get a 10% bonus on their top-up, and you earn a commission: 5% at the start, 10% after $1,000 in referral revenue, 15% after $2,000. Track earnings in your profile dashboard. Payouts are credited directly to your BananaBanana balance.

Yes — BananaBanana has a public remote MCP server at bananabanana.pro/mcp. AI agents and MCP clients such as Claude Code, Claude Desktop and Gemini CLI can generate images and video programmatically: create an API key in your profile, connect the server, and every generation is billed pay-as-you-go from the same balance you use on the website, at the same per-generation prices. A traditional REST API may follow later.

Yes — Claude and other AI agents can generate images, video and speech on BananaBanana through its public remote MCP (Model Context Protocol) server. Claude Code, Claude Desktop, claude.ai, Cursor, Gemini CLI, Codex, n8n and any other MCP-compatible client can connect. The agent gets the same models and settings as the web studio: reference images, first and last frames, Wan reference videos, Omni Extend up to 40 seconds, image and video editing, and Gemini 3.1 Flash TTS. Each call is billed from the same balance at the same prices, and video costs are quoted for confirmation before charging. The agent can then assemble the generated files on your machine with ffmpeg: BananaBanana handles generation, the agent handles the local edit. Sign in through OAuth from the client, or create an API key in Profile → MCP API Keys. Setup guide: bananabanana.pro/mcp.

BananaBanana offers Gemini 3.1 Flash TTS from Google, in the web studio and through MCP. It costs $0.01 per started 200 characters, so a 30-second line of about 450 characters is $0.03. There are 30 voices, a style prompt for tone and pace, and one narrator or a two-speaker dialogue in one file. Audio comes as a 24 kHz WAV file that you or your agent can lay over a video with ffmpeg.

Yes — with x402 an agent pays for each call in USDC on the Base network, without a BananaBanana account, API key or top-up. The agent requests a generation, receives the exact price, signs the payment and gets the result. It works for image, video and speech generation, and a call costs what the generation costs, from $0.01. If an x402 video fails, the payer receives a credit token for the same amount, valid for 90 days; payments are not refunded on-chain. Details: bananabanana.pro/mcp.

Yes — BananaBanana accepts card, PayPal and SEPA top-ups from $20. The amount you pay is added to your US-dollar balance without a bonus; volume bonuses of 5% from $50 and 10% from $100 apply to crypto top-ups only. Card payments happen inside PayPal's window, each top-up is a one-time payment with no subscription, and your balance never expires.