# Qwen API cost calculator

Alibaba Model Studio prices Qwen per token with separate input, output and cached-input rates. The prices here are the international (Singapore) list; the China region has its own list and model lineup. The context cache is implicit: a repeated prompt prefix is matched automatically and a hit is billed at 10.67%–20% of the input price. In our data Qwen3.8 Max, Qwen Max, Qwen3.7 Plus, Qwen Turbo list no cache-write fee, so a miss costs normal input, while Qwen3.8 Flash charge 1.33× the input price to write.

Some Qwen models have a Batch API at 50% off input and 50% off output for asynchronous jobs; models without a batch price are flagged in the table. Thinking mode is priced apart on Qwen Turbo: its thinking output costs 2.5× the standard output price, which the calculator applies when the output price switch is set to thinking. Long inputs cost more on tiered models: long-context rates apply above 256K for Qwen3.7 Plus and 2 more (input 1.25×–5×, output 1.25×–5×), to the whole request. 

The table prices one workload on each model: 10,000 requests a day with 6,000 input and 500 output tokens per request. Columns: no caching; 70% of input read from the cache and 5% written to it; every request sent through the Batch API. For the same traffic [Qwen3.8 Max](https://apicostcheck.com/models/qwen3.8-max/) costs 38× as much as [Qwen Turbo](https://apicostcheck.com/models/qwen-turbo/).

| Model | Standard 10,000 req/day · 6,000 in · 500 out [Calculate with your numbers →](https://apicostcheck.com/calculator/?daily=10000&input=6000&output=500&cache=0&write=0&batch=0) | With prompt caching 10,000 req/day · 6,000 in · 500 out · cache 70% [Calculate with your numbers →](https://apicostcheck.com/calculator/?daily=10000&input=6000&output=500&cache=70&write=5&batch=0) | Batch API 10,000 req/day · 6,000 in · 500 out · batch [Calculate with your numbers →](https://apicostcheck.com/calculator/?daily=10000&input=6000&output=500&cache=0&write=0&batch=100) |
| --- | --- | --- | --- |
| [Qwen3.8 Max](https://apicostcheck.com/models/qwen3.8-max/) | $4,500/mo | $2,295/mo | $4,500/mo no batch |
| [Qwen Max](https://apicostcheck.com/models/qwen-max/) | $3,840/mo | $2,227/mo | $1,920/mo |
| [Qwen3.8 Flash](https://apicostcheck.com/models/qwen3.8-flash/) | $340/mo | $176/mo | $340/mo no batch |
| [Qwen3.7 Plus](https://apicostcheck.com/models/qwen3.7-plus/) | $960/mo | $557/mo | $960/mo no batch |
| [Qwen Turbo](https://apicostcheck.com/models/qwen-turbo/) | $120/mo | $69.60/mo | $60.00/mo |

Estimates from published per-token rates; check official pricing before budgeting. Data synced from the LiteLLM price list 2026-10-12 12:00 UTC. 

## Cost per 1,000 requests

For 1,000 requests of 6,000 input and 500 output tokens, [Qwen Turbo](https://apicostcheck.com/models/qwen-turbo/) costs $0.40 at standard rates, $0.23 with caching and $0.20 through the Batch API; [Qwen3.8 Max](https://apicostcheck.com/models/qwen3.8-max/) costs $15.00, $7.65 and $15.00. Multiply by your own request count, or put your traffic into the calculator. 

## How to cut the bill, in order
- Put the system prompt, tools and documents first so the implicit cache hits on every request; only the user turn should change.
- Use Flash for routing and classification, Plus for chat and RAG, Max only for hard questions.
- Turn thinking mode off where it is not needed: on models with a separate thinking price the output tokens cost more, and reasoning tokens are billed as output.
- Keep requests under the tiered models' long-context threshold with retrieval; count prompts with the [token counter](https://apicostcheck.com/token-counter/).
- Send offline work through the Batch API where the model has one, and check the [caching calculator](https://apicostcheck.com/prompt-caching-calculator/) for your hit rate before budgeting in the [calculator](https://apicostcheck.com/calculator/). 

[When does caching pay off for your traffic? Use the prompt caching calculator →](https://apicostcheck.com/prompt-caching-calculator/)

## FAQ

### Does the Qwen context cache need code changes?

No. Prefix matching is automatic on supported models: keep the system prompt and documents at the start, and a hit is charged at the cached-input price. Model Studio also offers an explicit cache that you create and pay for by storage time; it is not in our data.

### Are China and international prices the same?

No. Model Studio keeps a separate price list and model lineup per region. This site shows the international list; an account on the China region should read the China price page.

### Which Qwen models charge more for long inputs?

Only the models named in the long-context sentence above; once a request passes the threshold the whole request is billed at the higher rate. The other models keep one price at any length in our data.

### Flash, Plus or Max?

Flash for high-volume classification, extraction and routing; Plus for most chat and RAG work; Max for hard reasoning and agents. The price gap between tiers is large, so route requests by difficulty and test two tiers on your own prompts.

---
Source: https://apicostcheck.com/qwen-api-cost-calculator/ · prices in USD per 1M tokens, synced 2026-10-12 12:00 UTC · Free to reuse with a link to https://apicostcheck.com (CC BY 4.0). Prices derived from LiteLLM's model price list, MIT License, (c) 2023 Berri AI.