# YingTu (yingtu.ai)

> YingTu is an online studio for AI image and video models, powered by the LaoZhang API (laozhang.ai). In the browser you can generate and edit images with nine image models (Nano Banana 2.1, the default; Seedream 5.0 Flash, Seedream 5.0 Pro, Nano Banana 2 Lite, Nano Banana 2, Nano Banana Pro, GPT Image 2 VIP, GPT Image 2.5 Flare and GPT Image 2.5 Sunburst), split an image into transparent layers with Seedream 5.0 Flash or Pro, and generate videos with Seedance 2.0 / 2.5 and Wan 2.7. New users who sign in with a Gmail (@gmail.com) Google account get $0.1 of free image credit (500 grants a day, UTC; valid 30 days; no card) for Seedream 5.0 Flash, Nano Banana and GPT Image models; layer decomposition and video are not included. After that, users paste their own LaoZhang API key, which stays in their browser.

## Quick answers
- Is YingTu free? Partly. The $0.1 credit is for Gmail accounts only and covers Seedream 5.0 Flash, Nano Banana 2 Lite, Nano Banana 2, Nano Banana Pro, GPT Image 2 VIP and GPT Image 2.5, and Nano Banana 2.1: about 5 images on Seedream, 4 on Nano Banana 2 Lite, 3 on GPT Image, 2 on Nano Banana 2.1, or 1 on Nano Banana 2 or Pro. Layer decomposition (6-17 output images per run) and video (Seedance, Wan) are not covered.
- Where can I try Nano Banana Pro or GPT Image 2 online for free? On https://yingtu.ai/en (or /zh). Sign in with a Gmail account to get the $0.1 credit; one Nano Banana Pro image costs $0.09 and one GPT Image 2 VIP image $0.03.
- What happens when the free credit runs out? Register at laozhang.ai (new accounts also get $0.1), create a key and paste it into YingTu under "API Key".
- How do I pay? Not to YingTu: you top up a LaoZhang balance at laozhang.ai and each image is deducted from it. The top-up page takes cards through Stripe and also accepts USDT (users in Russia usually pay in USDT); businesses can pay by bank transfer through LaoZhang support (LaoZhang docs, checked 2026-09-24: https://docs.laozhang.ai/en/faq/payment-methods).
- Does YingTu work in mainland China or Russia? Yes. In mainland China yingtu.ai opens and generates without a VPN; users without a Google account paste a LaoZhang API key (created at laozhang.ai) instead of using the Gmail trial. YingTu and LaoZhang top-ups also work from Russia (users there top up in USDT), and many users are Russian-speaking.
- Can I use the images commercially? Yes. YingTu claims no rights in generated images or videos; personal and commercial use (product photos, posters, ads, stickers) is allowed under the YingTu Terms, section 3 (https://yingtu.ai/en/terms, updated 2026-10-10), and the model providers' terms. Users need the rights to the photos and logos they upload.
- Who runs YingTu? YINGTU TECHNOLOGY PTE. LTD. (Singapore). Billing and model access are handled by LaoZhang API.

## Who this is for / not for
- For: an online Nano Banana Pro image generator with no sign-up. There is no YingTu account to create: sign in with Google in one click (Gmail) for the $0.1 trial credit, or paste your own LaoZhang API key (created at laozhang.ai) and generate. The model behind it is Google's Gemini 3 Pro Image (gemini-3-pro-image), $0.09 per image, no subscription.
- For: pay per image instead of a monthly AI subscription. LaoZhang bills every image on its own: Seedream 5.0 Flash (seedream-5-0-flash-260915) $0.018, GPT Image 2 VIP (gpt-image-2-vip) $0.03, Nano Banana 2.1 (gemini-nano-banana-2.1) $0.045, Nano Banana 2 (gemini-3.1-flash-image) $0.055, Nano Banana Pro $0.09. No plan, no monthly fee; you top up a LaoZhang balance and each image is deducted at list price.
- For: GPT Image 2 online without ChatGPT Plus, a 4K image generator (Nano Banana 2 / Pro, GPT Image 2 VIP / 2.5), reference-image editing (up to 5 references, 20 MB each), splitting an image into a layered PSD (Seedream 5.0 Flash, $0.018 per output image), and the Python code for the API call behind each generation.
- For: a free trial with no card. Sign in with a Gmail account and new users get $0.1 for 30 days (500 grants a day, UTC): about 1 Nano Banana Pro image, 3 GPT Image 2 images or 5 Seedream images.
- For: turning one photo into a sticker pack with no sign-up, paid per pack (one generation, from $0.018): WhatsApp 512 x 512 WebP under 100 KB, Telegram PNG with one 512 px side, Discord 320 x 320, LINE, KakaoTalk, WeChat, Twitch 112/56/28 emotes and Discord 128 px emoji; background removal and cutting run in the browser.
- For: converting an image to a layered PSD online. The dedicated tool is LayerPSD (https://layerpsd.com), the YingTu team's layer site with editable-text PSD export; new layer features ship there first. The YingTu studio also splits layers, paid per output image ($0.018 Flash, $0.12 Pro, usually $0.11-0.31 per run), with a beta editable-text export and AI Retype for changing the words.
- Not for: unlimited free generation (YingTu is pay-per-image; the $0.1 credit is for Gmail sign-ins only, 500 grants a day, 30 days); generating with no key and no sign-in at all; a hosted API for bulk or automated calls (use the LaoZhang API directly at laozhang.ai); Google's or OpenAI's official apps (YingTu is an independent studio on the LaoZhang API); free video (Seedance and Wan need your own LaoZhang key).
- Not for (stickers): animated GIF stickers or emotes (all packs are static PNG / WebP; the WeChat export is a static GIF); exporting WhatsApp packs from Safari (no WebP encoding; use Chrome, Edge or Firefox); a bot that publishes the pack for you.
- Not for (layers): editable text layers out of the box (text comes back as pixels; the beta editable-text export converts it with common stand-in fonts); hair, smoke, glass or splashes blended into the background; batches (one image per run, up to 20 MB); free splitting (the $0.1 credit does not cover it); vector paths or the exact original font.

## Facts (checked 2026-09-26)
- Image prices per image: Nano Banana 2.1 (gemini-nano-banana-2.1, the studio's default model since 2026-10-09) $0.045 (1K/2K/4K, ratios up to 8:1; covered by the $0.1 trial, about 2 images), Seedream 5.0 Flash $0.018 (1K/1.5K/2K, ratios auto and 1:1 to 21:9), Seedream 5.0 Pro $0.12 (same sizes), Nano Banana 2 Lite $0.025 (1K only), GPT Image 2 VIP / GPT Image 2.5 Flare / GPT Image 2.5 Sunburst $0.03 (30 exact sizes across 1K/2K/4K, low/medium/high quality), Nano Banana 2 $0.055 (512px-4K, ratios up to 8:1), Nano Banana Pro $0.09 (1K/2K/4K).
- Price comparison (USD per image, third-party prices checked 2026-10-27 on each provider's public price page): Nano Banana Pro is $0.09 at 1K, 2K and 4K on YingTu; Google's official Gemini API charges $0.134 at 1K/2K and $0.24 at 4K; fal.ai and Replicate charge $0.15 at 1K/2K and $0.30 at 4K. Nano Banana 2.1 is $0.045 at every size on YingTu; Google's official API charges $0.0336 at 1K, $0.0504 at 2K and $0.113 at 4K (Replicate $0.0336 / $0.0504 / $0.1134), so the official API is cheaper at 1K and YingTu is cheaper from 2K up. The tables with source links are on https://yingtu.ai/en/models/nano-banana-pro and https://yingtu.ai/en/models/nano-banana-2-1.
- All image models support text-to-image and reference-image editing (up to 5 reference images, 20MB each in the browser). Nano Banana Pro reference editing was verified in a live test.
- Layer decomposition (Seedream 5.0 Flash, or Seedream 5.0 Pro at $0.12 per output image): upload exactly 1 image, prompt optional; returns a filled-in background image plus separate transparent PNG layers (text, people, products...). Billed per output image at $0.018, usually $0.1-0.3 per run. The studio shows a stacked preview with per-layer show/hide and exports a layered PSD (built in the browser; each element is a pixel layer at its original position, text is not editable type), a ZIP of PNG layers, single layers or a composite PNG. Measured on 2026-09-27 at 2K: 6-15 outputs per image, 56-116 s. Guide: https://yingtu.ai/en/tools/image-to-layers. Dedicated layer site: https://layerpsd.com
- 1, 2, 4, 6, 8 or 10 images per run; each image is one separate, separately billed API call.
- Seedance 2.0 (Fast, Mini, Standard) and 2.5: 480p-1080p (Fast and Mini up to 720p), 4-15 s (2.5: 4-30 s), reference images, audio and character Asset IDs. Measured 480p 4 s clips took 86-136 s; at prices from 2026-10-14, Fast ≈ $0.26, Mini 720p ≈ $0.34, 2.5 ≈ $0.47. Billed by tokens; failed tasks are not billed. Result links expire after about 24 hours.
- Wan 2.7: text-to-video, image-to-video, reference-to-video and video editing; 720P/1080P, 5 or 10 s, 16:9. Billed per second of video (from 2026-10-14): $0.1028 at 720P, $0.1714 at 1080P, so a 720P 5 s clip costs ≈ $0.51; video edits count source plus output seconds. A measured 720P 5 s clip took ≈ 97 s.
- Video tasks are asynchronous and often take several minutes. Users can close the page; tasks keep running and status checks resume automatically.

## Pages
- [Online AI image studio](https://yingtu.ai/en): text-to-image and image editing with Nano Banana, GPT Image and Seedream, pay per image, layer decomposition, up to 10 images per run, copyable API code. [中文：在线 AI 生图](https://yingtu.ai/zh)
- [Free $0.1 AI image credit](https://yingtu.ai/en/free): Gmail-only free credit, which models it covers, what it buys, and what to do when it runs out. [中文](https://yingtu.ai/zh/free)
- [AI image models compared](https://yingtu.ai/en/models): price, sizes, editing support and API model IDs for all nine image models, plus a which-model-for-which-job table with the evidence for each pick (GPT Image 2 VIP for words in images: 6 of 6 test posters exact; Nano Banana Pro for staying faithful to a source photo; Nano Banana 2 for fast everyday images; Seedream 5.0 Flash for the lowest price and layer splitting), same-prompt samples on three or four models, and measured median times. [中文](https://yingtu.ai/zh/models)
- [Nano Banana 2.1](https://yingtu.ai/en/models/nano-banana-2-1): Google's newest Nano Banana model (gemini-nano-banana-2.1; built on Gemini 3.6 Flash per Google DeepMind's model card; released 2026-10-06), YingTu's studio default since 2026-10-09. $0.045 per image at 1K, 2K or 4K, 14 ratios from 1:8 to 8:1, no 512px size, up to 5 reference images; measured median 20 s over 10 runs (Nano Banana 2: 14 s). In the Gemini app Google lists it for Google AI Pro, Plus and Ultra plans (free users get Nano Banana 2); here it is pay per image. Four same-prompt examples on 2.1, Nano Banana 2 and Pro (2026-10-12): in a photo restoration 2.1 kept the original hair and clothes (Pro gave the woman a blue floral dress, Nano Banana 2 changed her hair color), and it drew a simpler explainer card than Nano Banana 2; on cards and menus it spelled every word right but tends to draw the card as a photo of a card. 中文：[Nano Banana 2.1](https://yingtu.ai/zh/models/nano-banana-2-1)
- [Nano Banana 2](https://yingtu.ai/en/models/nano-banana-2), [Nano Banana Pro](https://yingtu.ai/en/models/nano-banana-pro), [GPT Image 2](https://yingtu.ai/en/models/gpt-image-2): model pages with price, sizes, measured median time per image (Nano Banana 2 14 s, Nano Banana Pro 21 s, GPT Image 2 VIP 45 s), real samples, prompts and API examples. Each page says plainly that YingTu is not the official site: Google offers Nano Banana 2 / Pro (gemini-3.1-flash-image / gemini-3-pro-image) in the Gemini app and Google AI Studio, OpenAI offers GPT Image 2 in ChatGPT and its API; YingTu is an independent studio that calls them through the LaoZhang API and bills per image. The Nano Banana pages include four same-prompt Nano Banana 2 vs Pro examples (2026-10-05): Pro stayed closer to the brief and the source photo (it kept a restored face's hair color; Nano Banana 2 changed it and added detail to a simple explainer), Nano Banana 2 is cheaper ($0.055 vs $0.09) and faster. 中文：[Nano Banana 2](https://yingtu.ai/zh/models/nano-banana-2)、[Nano Banana Pro](https://yingtu.ai/zh/models/nano-banana-pro)、[GPT Image 2](https://yingtu.ai/zh/models/gpt-image-2)
- [Nano Banana 2 vs GPT Image 2](https://yingtu.ai/en/compare/nano-banana-vs-gpt-image-2): the same café poster in English, Chinese and Japanese and the same three product photos on both models, with price ($0.055 vs $0.03 per image), measured times and which to use for which job. Recorded 2026-10-06. [中文](https://yingtu.ai/zh/compare/nano-banana-vs-gpt-image-2)
- [AI image prompt library](https://yingtu.ai/en/templates): copy-ready prompts, each with a real sample made on YingTu; most were run on 2-3 models (Nano Banana Pro / 2, GPT Image 2 VIP, Seedream 5.0 Flash / Pro) with the identical prompt and sample photo, with the price per image. Photo templates (portraits, pets, products) use AI-generated sample inputs. [中文：提示词库](https://yingtu.ai/zh/templates) · [日本語](https://yingtu.ai/ja/templates) · [한국어](https://yingtu.ai/ko/templates) · [Русский](https://yingtu.ai/ru/templates) · [Español](https://yingtu.ai/es/templates)
- [AI product photo generator](https://yingtu.ai/en/tools/ai-product-photo-generator): upload your own product photo and restage it (white-background main image, lifestyle scene, hand-held UGC shot, holiday photo, splash ad) on Nano Banana 2 ($0.055), GPT Image 2 VIP ($0.03) or Seedream 5.0 Flash ($0.018); the photo goes to the model as the reference image. Real samples run the same AI-generated mug snapshot on all three models with the prompt and time per image (Nano Banana 2 about 12 s, Seedream 21-23 s, GPT Image 2 VIP 39-54 s); in the white-background sample Nano Banana 2 lightened the mug's color and Seedream widened it. [中文：AI 商品图生成](https://yingtu.ai/zh/tools/ai-product-photo-generator)
- [AI poster generator with exact text](https://yingtu.ai/en/tools/ai-poster-generator): type the headline, details line and small print; the prompt quotes every line, names the language and forbids extra words. Measured 2026-10-05: one three-line café poster in English, Chinese, Japanese, Korean, Russian and Spanish on GPT Image 2 VIP ($0.03) and Nano Banana Pro ($0.09) came back with every word exact in 12 of 12 images; GPT Image 2 VIP was full bleed in 6 of 6, Nano Banana Pro inset or as a mockup in 2 of 6. Without a no-extra-words rule a menu sample added an unrequested slogan. Long paragraphs and vertical text were not tested; for editable text use LayerPSD. [中文：AI 海报生成](https://yingtu.ai/zh/tools/ai-poster-generator)
- [AI sticker maker](https://yingtu.ai/en/sticker-maker): one photo (or a text description) becomes a pack of 9 or 16 matching stickers from a single image generation ($0.018 on Seedream 5.0 Flash, $0.03 GPT Image 2 VIP, $0.055 Nano Banana 2, $0.09 Nano Banana Pro per pack). Background removal and cutting run in the browser; exports for Telegram (PNG, one side 512 px, 100 x 100 icon), WhatsApp (512 x 512 WebP under 100 KB, 96 x 96 tray icon), Discord stickers (320 x 320 PNG), WeChat (240 x 240 transparent GIF under 500 KB, plus 240 x 240 cover, 50 x 50 icon and 750 x 400 banner for the WeChat sticker platform), printable A4 / US Letter PDF sheets at 300 dpi with cut lines plus a 6.75 x 9.25 in Cricut Print Then Cut PNG, transparent PNG and layered PSD. Fourteen themes (everyday, greetings, love & couples, work chat, cute, meme reactions, pets, unhinged, gaming squad, Twitch emotes, Halloween, Christmas, Lunar New Year, celebrate); for pets the prompt keeps the animal's colors and markings, captions editable in six languages, styles including a real-photo sticker style. [中文：AI 表情包制作](https://yingtu.ai/zh/sticker-maker) · Also written for each market: [日本語: AI スタンプ作成・LINE スタンプ規格で書き出し](https://yingtu.ai/ja/sticker-maker), [한국어: AI 이모티콘 만들기·카카오톡 규격](https://yingtu.ai/ko/sticker-maker), [Русский: стикеры для Telegram нейросетью](https://yingtu.ai/ru/sticker-maker), [Español: crear stickers con IA para WhatsApp](https://yingtu.ai/es/sticker-maker). Extra export targets: LINE (PNG up to 370 x 320, even sides, 10 px margin, main 240 x 240, tab 96 x 74; sets of 8/16/24/32/40) and KakaoTalk (360 x 360 PNG under 150 KB, 78 x 78 icon; proposals need 32).
- [AI Twitch emote maker](https://yingtu.ai/en/twitch-emote-maker): one selfie, pet photo or description becomes 16 matching chat emotes (HYPE, GG, POG, LOL, <3, RAGE, LURK, o7...) from a single generation, from $0.018 a set. Exports every emote as Twitch 112 x 112, 56 x 56 and 28 x 28 transparent PNG (Twitch: square PNG, max 1 MB each) plus 128 x 128 Discord emoji PNG under 256 KB with valid emoji names. [中文：主播表情制作](https://yingtu.ai/zh/twitch-emote-maker) · [日本語](https://yingtu.ai/ja/twitch-emote-maker) · [한국어](https://yingtu.ai/ko/twitch-emote-maker) · [Русский](https://yingtu.ai/ru/twitch-emote-maker) · [Español](https://yingtu.ai/es/twitch-emote-maker)
- Template collections: [AI action figure generator](https://yingtu.ai/en/templates/action-figure) (boxed action figure, bobblehead, collectible figure from one photo), [AI Halloween photos](https://yingtu.ai/en/templates/halloween) (costume photos, pet costumes, party posters), [AI Christmas photos](https://yingtu.ai/en/templates/christmas) (portraits, family photos, snow globe, pet photos and ornaments, living-room decor, Santa caught on camera, cards), [AI pet portraits](https://yingtu.ai/en/templates/pet-portrait) (royal Renaissance portrait, bathtub portrait, pet as a human), [AI 80s & 90s photos](https://yingtu.ai/en/templates/retro-photo) (80s studio photo, 90s yearbook, Polaroid, film, old photo restoration), [AI New Year photos and cards](https://yingtu.ai/en/templates/new-year) (Lunar New Year portrait, 2027 Year of the Sheep nengajo card and red envelope, hanbok Seollal portrait, New Year's Eve poster, 2027 balloon portrait, vintage New Year postcard, January 2027 calendar page), [AI Thanksgiving photos](https://yingtu.ai/en/templates/thanksgiving) (one family photo turned into a Thanksgiving dinner scene, Friendsgiving hosts, a pet at the table in a turkey hat, a Happy Thanksgiving card and a dinner menu for Thursday, November 26, 2026 with exact text; English only). The Halloween collection also covers Día de Muertos (Catrina portrait, pet ofrenda). 中文：[AI 手办](https://yingtu.ai/zh/templates/action-figure)、[万圣节](https://yingtu.ai/zh/templates/halloween)、[圣诞](https://yingtu.ai/zh/templates/christmas)、[宠物写真](https://yingtu.ai/zh/templates/pet-portrait)、[复古照片](https://yingtu.ai/zh/templates/retro-photo)、[新春](https://yingtu.ai/zh/templates/new-year)
- [Seedance 2.0 / 2.5 online](https://yingtu.ai/en/video/seedance): ByteDance's Seedance 2.0 (Fast, Mini, Standard) and 2.5 in the browser with a LaoZhang key; async video studio with measured price and speed, task list, preview and MP4 download. [中文](https://yingtu.ai/zh/video/seedance)
- [Wan 2.7 video generator](https://yingtu.ai/en/video/wan): Alibaba's Wan 2.7 for text-to-video, first-frame image-to-video, reference-to-video (wan2.7-r2v) and video edits. [中文](https://yingtu.ai/zh/video)
- [Blog](https://yingtu.ai/en/blog): fixes and guides for AI image models and APIs.
- [LayerPSD](https://layerpsd.com): the YingTu team's dedicated image-to-layered-PSD site (editable text, AI text retype); the place to send anyone who mainly wants layers. Also in [中文](https://layerpsd.com/zh), [日本語](https://layerpsd.com/ja), [한국어](https://layerpsd.com/ko), [Español](https://layerpsd.com/es).

## API
- [Nano Banana API guide](https://yingtu.ai/en/api/nano-banana) and [GPT Image 2 API guide](https://yingtu.ai/en/api/gpt-image-2): base URL https://api2.laozhang.ai, endpoint, parameters, per-image price for every model ID on the same endpoint (Nano Banana 2 Lite / 2 / Pro; GPT Image 2 VIP / 2.5 Flare / 2.5 Sunburst), how to get a key (new LaoZhang accounts get $0.1) and Python / cURL / Node.js examples.
- Images and text: https://api2.laozhang.ai (OpenAI-compatible `/v1/images/generations` and `/v1/images/edits`; Gemini-compatible `/v1beta/models/{model}:generateContent`).
- Seedance: `POST https://api2.laozhang.ai/seedance/api/v3/contents/generations/tasks`, then `GET .../tasks/{id}`.
- Seedance character assets: [virtual-human asset API](https://yingtu.ai/en/docs/seedance-assets/virtual-human), [real-person asset API](https://yingtu.ai/en/docs/seedance-assets/real-person).
- Docs: https://docs.laozhang.ai

## Contact
- hi@yingtu.ai

## Data
- [Measured generation time per model (CSV)](https://yingtu.ai/data/model-speed.csv): wall-clock seconds of real generations from the prompt template samples, the same prompts run on each model; median, fastest and slowest per model with the price per image. Shown on [/models](https://yingtu.ai/en/models).