Cycling through all six. Tap any point to stop.

Measured on six things

Nano Banana, GPT Image & Seedance AI image and video

Speed

Lighthouse mobile 51/100 with a rough CLS of 0.349 and 1.1s TBT — the layout shifts under you while you're trying to click generate. LCP 2.0s is fine; the jitter is the problem.

Screenshots

1 / 2

Nano Banana CN

About this app

Nano Banana CN is a Chinese-friendly AI image and video studio. Use Nano Banana Pro, Nano Banana 2 and 2.1, GPT Image 2, GPT Image 2.5, Seedance 2.5, Seedance 2.0 and more for text-to-image, image-to-image, text-to-video, image-to-video and video-to-video. One-time credits never expire, and an API is available for image models. Nano Banana CN is an independent site and is not affiliated with Google.

Key features

1Multi-model hub: Nano Banana Pro/2/2.1, GPT Image 2/2.5, Seedance 2.5/2.0, Seedream, Tongyi — text/image/video generation in one place.
2Chinese-first UX with localized prompts, browser-language detection, and Alipay checkout for the domestic market.
3Credit-based pricing with permanent validity above ¥999, plus an image-model API for developers.

Target customer profile

Drawn from the product itself, not from a survey.

Scout Review

Written by AI from measured evidence, scored out of 100.

Nano Banana CN is a competent, clearly-positioned Chinese-language AI image and video studio that aggregates Google's Nano Banana models alongside GPT Image, Seedance and others. The composer-first homepage and before/after gallery sell the product honestly, and the changelog shows a team shipping real improvements weekly. The weaknesses are concrete: a layout that shifts under your cursor (CLS 0.349), a security posture missing most modern headers and DMARC, and a primary action styled so faintly it fails contrast. It's a solid mid-tier tool with a genuine localization wedge — not a category leader, but not pretending to be.

Measure by measure

Usability

Clear prompt-to-generate path above the fold, cost shown per generation, and a changelog full of real UX fixes. Abbreviated toggles and hidden pricing styling add friction for newcomers.

Design

Coherent rounded-card system and a composer-first hero that sells the tool over marketing copy. Undermined by a wrapping toolbar, near-invisible send button, and three indistinguishable model cards.

Speed

Lighthouse mobile 51/100 with a rough CLS of 0.349 and 1.1s TBT — the layout shifts under you while you're trying to click generate. LCP 2.0s is fine; the jitter is the problem.

Security

HTTPS and HSTS in place, SPF set, and legal pages linked. But 5/13 header checks, no CSP, no frame protection, no secure cookies, and no DMARC — thin for a site handling payments.

Accessibility

Lighthouse a11y 92/100 and genuine multi-language support are real strengths. Failing contrast and unnamed buttons — matching the low-contrast send arrow — keep it from the top of the band.

Growth

A real, specific wedge: Chinese-language users of Google's image models, with localized prompts, Alipay, and a multi-model hub plus API. Differentiated, but reseller-shaped with no moat.

Improvement areas

Layout stability (CLS 0.349) and missing security headers (CSP, frame protection, DMARC) on a payment-handling site.
Low-contrast primary action and credit cost, plus unnamed icon buttons failing accessibility audits.
Indistinguishable model cards and tooltip-less abbreviated toggles that force new users to guess.

Design & Usability

The homepage leads with the tool, not a wall of marketing — the composer card puts prompt, reference, model, ratio and cost in one place, and the before/after carousel shows actual output. That's the right instinct for an AI generator. The execution has gaps: the toolbar wraps awkwardly, the send arrow and '2/张' cost are styled in near-invisible gray, and the three model cards are so similar you can't tell Nano Banana 2 from Pro from 2.1 without reading. Usability is genuinely decent — the path from prompt to generate is obvious, cost is disclosed per generation, and the changelog shows real iteration (interactive editor, drag-drop uploads, ticket support). The friction is in the details: abbreviated toggles with no tooltips force guessing.

Speed & Security

Speed is the weak spot. Lighthouse mobile scores 51/100, and while LCP at 2.0s is acceptable, a CLS of 0.349 means the layout visibly shifts as it loads — bad when the thing you're trying to click is a generate button. TBT of 1.1s adds lag to interaction. The site claims 5s average generation, but that's a marketing number, not a measured one. Security is thin for a payment-handling site: HTTPS and HSTS are enforced and SPF is set, but only 5 of 13 header checks pass — no CSP, no frame protection, no X-Content-Type-Options, no Referrer-Policy, no secure cookies, and no DMARC. Legal pages (terms, privacy, refund) are linked and the changelog mentions API key auth hardening, but there's no published audit, SOC 2, or bug bounty to point to.

Accessibility & Growth

Accessibility scores well on the automated audit (92/100) and the product genuinely supports multiple languages with browser detection — a real plus for its Chinese-first audience. But the failing audits are telling: buttons without accessible names and insufficient contrast, which line up exactly with the low-contrast send arrow and gray credit cost seen in the visual audit. On growth, the idea has a real wedge: Chinese-speaking users of Google's Nano Banana models who want localized prompts, Alipay checkout, and one hub for many models (Nano Banana, GPT Image, Seedance, Seedream, Tongyi) plus an image API. That's a specific, underserved segment. The limit is structural — it's an independent aggregator with no moat, dependent on upstream models it doesn't control, and the site itself has to disclaim Google affiliation.

Conclusion

If you're a Chinese-speaking creator who wants Nano Banana output without wrestling an English UI or a foreign payment flow, this does the job and the credit pricing is transparent. If you're evaluating it for anything payment-sensitive or accessibility-critical, the missing headers and unnamed buttons are worth flagging to the team. The bones are good and the iteration cadence is real — the next few months of changelog will tell you whether it matures into something durable or stays a storefront.

Suggestions for improvement

Fix the CLS 0.349 layout shift — reserve space for the composer and gallery so the page doesn't jump on load.
Add the missing security headers (CSP, X-Frame-Options, X-Content-Type-Options, Referrer-Policy) and publish a DMARC record.
Raise contrast on the '2/张' credit cost and send button, and give icon-only buttons accessible names.
Differentiate the three model cards with badges or sample thumbnails, and add tooltips to the '加速' and ratio toggles.

How we score

Alternatives comparison

Named competitors, point by point. Nobody paid to appear here or to be left out.

	Nano Banana CN
Chinese-language interface and localized prompts	Yes — Chinese-first with browser-language detection	Not specified in provided details
Number of AI image/video models aggregated	Many — Nano Banana, GPT Image, Seedance, Seedream, Tongyi	Not specified in provided details
Local payment support (Alipay)	Yes — Alipay supported	Not specified in provided details
Credit validity / pricing model	Credits permanent above ¥999; tiers from ¥58	Not specified in provided details
Developer API for image models	Yes — image model API available	Not specified in provided details