China LLM pricing
China LLM API Pricing Calculator
Compare 中国大模型 API 成本 in CNY across Qwen, Doubao, ERNIE, Hunyuan, and your own negotiated contract price. Use the same request volume and token profile before choosing a model or setting SaaS margins.
中国大模型 API 价格比较
| 厂商 | 模型 | 输入 / 百万 | 输出 / 百万 | 折扣 | 月成本 | 单次请求成本 |
|---|
Keep currency clean
Compare Chinese API prices in yuan first instead of mixing them into USD tables.
Model private quotes
Add a negotiated input and output price when a provider gives you a lower enterprise rate.
Start from real workloads
Switch between customer support, RAG, agent workflow, and content generation assumptions.
How to use this China LLM API price comparison
Start with monthly API requests, average input tokens, and average output tokens. The calculator ranks models by estimated monthly CNY cost. If you have a vendor discount, enter it as a percentage. If you have exact contract numbers, turn on the custom row and enter your own input and output price per million tokens.
This is useful when you are comparing 阿里云百炼 Qwen, 火山引擎 Doubao, 百度千帆 ERNIE, 腾讯混元 Hunyuan, and an overseas API shortlist for the same product. Do not mix CNY and USD list prices until you have a final exchange-rate assumption.
If you already know the provider shortlist, start with a focused page: Qwen API Cost Calculator for Alibaba Cloud Bailian or Doubao API Cost Calculator for Volcano Engine.
| Decision | Use this input | Watch out for |
|---|---|---|
| Which China model is cheapest? | Same requests, input tokens, and output tokens across every provider. | Endpoint version, context length, and output-heavy workloads. |
| Is a vendor quote actually better? | Custom contract input and output price per 1M tokens. | Minimum spend, resource-package expiry, and included quota rules. |
| Can this support a paid product? | Monthly cost plus budget buffer. | Retries, evaluation runs, logs, moderation, and support cost. |
Self-audit: current coverage gaps
This page currently prioritizes models with official CNY pricing that can be compared cleanly. DeepSeek, Kimi, GLM, and MiniMax have strong developer demand, but their public API prices are often published in USD, split by cache tier, long-context tier, or plan type. They should be added only after their currency and billing rules are modeled explicitly.
That is a deliberate quality rule: a smaller verified calculator is better than a larger table that mixes CNY list prices, USD endpoint prices, promotional package prices, and unofficial reseller rates without warning.
Customer buying checklist
- Ask whether the quote is pay-as-you-go, prepaid package, or minimum-spend contract.
- Check whether input, cached input, output, tool calls, and search calls are billed separately.
- Confirm whether the price changes for long context, batch inference, priority traffic, or region.
- Compare the quoted price against your own workload, not only the provider's headline rate.
- Keep a launch buffer for retries, evaluation runs, logging, moderation, and support workflows.
FAQ
Should China LLM API prices be compared separately from USD models?
Yes. Chinese providers usually publish CNY prices and may use resource packages, postpaid rates, model-specific tiers, or negotiated discounts. Compare them in CNY first, then convert only when you need a final business budget.
How should negotiated model discounts be handled?
Use the contract discount field for a broad percentage discount, or add a custom contract price row when your vendor gives you a specific input and output token price.
Why are some Chinese AI models not listed yet?
Some providers publish prices in USD, some publish CNY, and some use package or promotion pricing. This page keeps the CNY comparison separate until each model has a clear official price source and currency treatment.
Sources
China model prices in this calculator were last checked on 2026-07-04. Pricing can change without notice.