Model Price Comparison

Yes. We purchase in bulk from model providers and operate our own computing centers (green computing) in Zhongwei, Ningxia and Ulanqab and Horinger, Inner Mongolia, so our aggregated prices are lower than buying directly from each provider's official channels. The "Official" and "Our Site" columns in the table let you compare item by item.

Identical prompts are shared across users through a common cache. Once a cache hit occurs, the input portion is no longer recomputed, cutting the price by more than 90%. Our lowest cache-hit price is ¥0.010 per million Tokens. Models marked "Not yet supported" in that column currently have no cache pricing.

There is no monthly fee and no minimum spend. Everything is billed by actual token usage, so you pay only for what you use. After a top-up, the balance is credited in real time, and you can view usage details anytime in the console.

You can create multiple sub-keys under one main account and set quotas for each. Members use their own keys with separately managed quotas, while billing is viewed centrally under the main account.

Prices are periodically synced from each vendor's official pricing page. The comparison table lists input price, output price, cache hit price, and context window, allowing item-by-item comparison between official direct-purchase prices and our aggregated prices.

Sorted by input price, the lowest tier currently is the Flash version of the Qwen series. The comparison table supports sorting by price, and each model lists both the official direct-purchase price and our aggregated price, making the difference clear at a glance.