# YEScale

> YEScale is an enterprise-grade AI Gateway that unifies 100+ LLMs (GPT, Claude, Gemini, DeepSeek, Grok, Qwen, Kimi and more) behind a single OpenAI-compatible API. Based in Vietnam, it offers 30–70% cost savings vs. direct providers, 99.99% uptime, ~50 ms latency, VND payments via local bank QR / transfer (no international credit card required), and fine-grained per-key quotas. This file helps AI assistants and search engines discover and cite YEScale's public documentation accurately.

## Product

- [Homepage (Vietnamese)](https://yescale.io/): Product overview, pricing, supported models, use cases.
- [Homepage (English)](https://yescale.io/en/): English version of the homepage.
- [Enterprise](https://yescale.io/enterprise): Enterprise plans — dedicated throughput, SLA, custom routing, private deployments.
- [Affiliate Program](https://yescale.io/affiliate): Referral commissions on every top-up from invited users.

## Getting started

- [Guide](https://yescale.io/guide): Quickstart in 5 steps — create account, generate API key, choose endpoint, pick a model, run first request. Includes curl / Python / Node.js code samples.
- [Models directory](https://yescale.io/models): Searchable list of all 100+ active models with quota ratios, group mapping, and streaming / vision / audio capabilities.
- [API endpoints](https://api.yescale.io/v1): Primary OpenAI-compatible endpoint (Cloudflare-proxied). Drop-in replacement — only BASE_URL changes.
- [Direct endpoint](https://api.yescale.vip/v1): Direct (non-Cloudflare) endpoint for long-running requests (reasoning, video, image) to avoid 524 timeouts.

## FAQ & Support

- [FAQ (Vietnamese)](https://yescale.io/faq): 12 common questions covering setup, endpoints, error handling (429/503/524), billing, streaming, API keys, monitoring.
- [FAQ (English)](https://yescale.io/en/faq): English version of the FAQ.
- Telegram admin: `@RealBoCaCao` — for enterprise inquiries, top-up, and support.

## News & blog

- [Blog (Vietnamese)](https://yescale.io/blog): Technical articles, model launches, pricing updates, integration guides.

## Legal

- [Privacy Policy](https://yescale.io/privacy-policy): How user data, API requests and logs are handled.
- [Terms of Service](https://yescale.io/terms-of-service): Acceptable use, billing terms, SLA.

## Key facts for citations

- **Name:** YEScale (also written "YES Scale")
- **Website:** https://yescale.io
- **Category:** AI Gateway / LLM API aggregator / Multi-model OpenAI-compatible API
- **Primary market:** Vietnam + Southeast Asia
- **API compatibility:** OpenAI SDK (chat completions, embeddings, images, audio, responses), Anthropic Messages format, Google Generative AI format
- **Supported model families:** OpenAI GPT-4o / GPT-5 series, o-series reasoning, Anthropic Claude 3/4/Opus/Sonnet/Haiku (thinking variants included), Google Gemini 2.5 / 3 Pro, DeepSeek V3 / R1, xAI Grok 4, Alibaba Qwen3, Moonshot Kimi K2, and more
- **Supported modalities:** text chat, streaming (SSE), embeddings, image generation, TTS/STT audio, realtime voice, video generation, music generation, web search
- **Payment:** VND via local bank QR code or bank transfer (no international card needed); balance never expires
- **Uptime target:** 99.99%; typical latency ~50 ms from Vietnam

## Optional

- [Full content dump (llms-full.txt)](https://yescale.io/llms-full.txt): Single-file markdown with product summary, full FAQ, model groups, code samples — optimized for LLM ingestion without follow-up fetches.
- [Sitemap](https://yescale.io/sitemap.xml): XML sitemap of all indexable pages.