Cycling through all six. Tap any point to stop.

Measured on six things

OpenAI-compatible API gateway for Chinese LLMs. One key for Qwen, DeepSeek, Kimi, GLM and more. USD billing via Stripe. Just change your base URL.

Usability

The pitch is refreshingly simple — swap your base URL, keep your SDK — and nav is clean. But there's no visible console link, no language switch for a bilingual audience, and no demo or trial path beyond 'Get API Key'.

Screenshots

1 / 1

SinoGate

About this app

SinoGate is an OpenAI-compatible API gateway that gives overseas developers one key to Chinese foundation models: Qwen, DeepSeek, Kimi, GLM, MiniMax and more. Swap your base URL and keep your existing SDK. Pay in USD with Stripe, and set up failover channels so requests keep flowing.

Key features

1One OpenAI-compatible API key and base URL for Qwen, DeepSeek, Kimi, GLM, MiniMax and more.
2USD billing via Stripe — no local-currency payment juggling per Chinese provider.
3Failover channels across model providers so requests keep flowing during an outage.

Target customer profile

Drawn from the product itself, not from a survey.

Demographic

Overseas developers using OpenAI-compatible SDKs

Pain points

Separate accounts and base URLs for each Chinese model provider

Primary needs

One API key and a base URL swap for Qwen, DeepSeek, Kimi, GLM and more

Demographic

Teams that want to pay in USD

Pain points

Paying Chinese model providers through local currencies and payment methods

Primary needs

USD billing via Stripe

Demographic

Products that need requests to keep flowing

Pain points

A single provider outage stops their AI features

Primary needs

Failover channels across model providers

Scout Review

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

SinoGate is a well-executed early product with a genuinely useful pitch: one OpenAI-compatible key for Qwen, DeepSeek, Kimi, GLM and MiniMax, USD billing via Stripe, and failover across providers. The site is fast (87 mobile / 92 desktop Lighthouse), the header security is strong (11/12), and accessibility passes automated checks at 100. The weaknesses are consistent: the design sells atmosphere over product (no code, endpoint or model list in the hero), the trust layer is thin for a key-holding proxy (no SOC 2, pen-test or bug bounty), and the growth story rests on a copyable base-URL swap. It's a competent, clearly positioned tool that needs to show its work — literally — to convert skeptical developers.

Measure by measure

Usability

Design

Committed, coherent art direction — serif wordmark, gold CTA, painted hero — that reads premium. But it sells a mood, not a developer tool: no code, endpoint or model list in the hero, and the Orientalist imagery is a theme, not a signal.

Speed

Lighthouse mobile 87/100 and desktop 92/100, with CLS 0 and TBT 23ms — a stable, snappy page. The one soft spot is a 3.1s mobile LCP, likely the heavy hero image doing the lifting.

Security

Strong header hygiene: HTTPS enforced, HSTS, CSP, frame protection, Referrer-Policy, SPF and DMARC — 11/12 on the probe, missing only Permissions-Policy. No published SOC 2, pen-test or bug-bounty page found, which is the gap for a product holding API keys.

Accessibility

Lighthouse accessibility 100/100 on automated checks — a genuinely clean baseline. The caveat is manual: the visual audit caught tiny, low-contrast micro-labels the axe run can't see, and there's no language switcher for a bilingual audience.

Growth

A real, specific pain — juggling five Chinese providers, local-currency payments and outages — with a concrete wedge: one key, USD Stripe billing, failover. Early (domain 0.3 yrs), and gateways are a copyable layer.

Improvement areas

No published security/trust page (SOC 2, pen-test, bug bounty) for a key-holding proxy.
Hero shows atmosphere, not product — no code snippet, endpoint or model list.
No visible Console link or language switcher; low-contrast micro-labels hurt readability.

Design & Usability

SinoGate's landing page has real craft: a high-contrast serif wordmark, a single gold CTA that owns the fold, and a painted hero with a committed mood. The problem is that the mood is 'mystic East' rather than 'developer tool' — there's no code snippet, no endpoint, no model list anywhere in the hero, so the actual product is invisible. Usability follows the same pattern: the value prop (swap your base URL, keep your SDK) is genuinely simple and the nav is clean, but there's no console link, no language switch for a bilingual audience, and no demo or trial path. You either click 'Get API Key' or you leave. For a gateway whose whole job is to be a drop-in replacement, showing the drop-in would beat showing a sunset.

Speed & Security

Speed is a genuine strength. Lighthouse puts mobile at 87/100 and desktop at 92/100, with CLS 0 and TBT 23ms — a stable, non-janky page. The only soft metric is a 3.1s mobile LCP, almost certainly the full-bleed hero image, which is a fair trade for the visual identity but worth compressing. Security is solid on the technical layer: HTTPS enforced, HSTS, CSP, frame protection, Referrer-Policy, SPF and DMARC all present, 11/12 on the header probe with only Permissions-Policy missing. What's absent is the trust layer a key-holding proxy actually needs — no SOC 2, no pen-test summary, no bug-bounty page, no data-handling detail beyond a privacy policy. For a product that sits between your app and third-party model providers, that's the gap to close next.

Accessibility & Growth

Accessibility scores 100/100 on Lighthouse's automated axe-core run, which is a clean baseline and better than most indie tools manage. The caveat is manual: the visual audit caught micro-labels like 'GLOBAL ACCESS' rendered too small and low-contrast to read, and there's no language switcher despite an explicitly bilingual audience. Automated checks are a floor, not a ceiling. On growth, the idea holds up: overseas developers genuinely suffer from five separate Chinese-provider accounts, local-currency payments and single-provider outages, and SinoGate's wedge — one key, USD Stripe billing, failover channels — is concrete and demonstrable. Against OpenAI API (no Chinese models) and Alibaba Cloud Model Studio (single-vendor), it's differentiated. The honest limit: a base-URL proxy is a thin, copyable layer, and the domain is only 0.3 years old, so the moat has to be built from reliability and trust, not the swap itself.

Conclusion

If you're an overseas developer tired of maintaining five Chinese-provider accounts and paying in five currencies, SinoGate is aimed squarely at you and the value prop is real. Before committing production traffic, ask for the security posture in writing — key handling, logging, retention — since none of it is published. And the team should put a code block in the hero: a curl call, a base URL, a model list. That single change would do more for conversion than any painted gate. The foundation is solid; the next move is showing the product instead of the scenery.

Suggestions for improvement

Put a real code snippet and the base URL in the hero — show the drop-in swap, not just describe it.
Publish a security/trust page: key handling, logging, retention, and ideally a pen-test summary or bug bounty.
Add a visible Console link and a language switcher for the bilingual overseas-developer audience.
Fix the low-contrast micro-labels and compress the hero image to pull mobile LCP under 2.5s.

How we score

Alternatives comparison

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

	SinoGate	OpenAI API	Alibaba Cloud Model Studio
Access to Chinese foundation models (Qwen, DeepSeek, Kimi, GLM, MiniMax)	Yes — multi-provider gateway	No — OpenAI models only	Partial — Qwen and Alibaba-hosted models
OpenAI-compatible SDK / base-URL swap	Yes — core pitch	Native (it is OpenAI)	Varies by endpoint; not a drop-in swap
USD billing via Stripe	Yes — Stripe in USD	Yes — USD	Alibaba Cloud billing; local-currency friction
Cross-provider failover channels	Yes — failover across providers	No cross-provider failover	Within Alibaba ecosystem only
Published security certifications (SOC 2 / pen-test)	Not found in collected evidence	Yes — extensive compliance program	Yes — enterprise cloud compliance