Alibaba's Qwen API models, the Qwen LLM family, on APIMart: Qwen3.8-Max, Qwen3.8-Flash, Qwen3.8-27B, Qwen3.7-Plus, Qwen3.7-Flash and more, through OpenAI-compatible Chat Completions or the Responses API with built-in tools, on one API key.
Service Highlights
Guides, comparisons and pricing analysis from the APIMart blog.
What's on your mind?
Ask APIMart
qwen4-max
Transparent pricing with no hidden fees. Pay only for what you use.
Per 1M tokens
| Rate | GoldCurrent price | Platinum | Diamond | Official | Your savings |
|---|---|---|---|---|---|
| Input | 14.2857~$1.4286 | 13.5714~$1.3571 | 12.8571~$1.2857 | 17.8571~$1.7857 | 20% |
| Cached input | 0.7143~$0.0714 | 0.6786~$0.0679 | 0.6429~$0.0643 | 0.8929~$0.0893 | 20% |
| Cache write | 17.8571~$1.7857 | 16.9643~$1.6964 | 16.0714~$1.6071 | 22.3214~$2.2321 | 20% |
| Cache write (5 min) | 17.8571~$1.7857 | 16.9643~$1.6964 | 16.0714~$1.6071 | 22.3214~$2.2321 | 20% |
| Cache write (1 hr) | 28.5714~$2.8571 | 27.1429~$2.7143 | 25.7143~$2.5714 | 35.7143~$3.5714 | 20% |
| Output | 42.8571~$4.2857 | 40.7143~$4.0714 | 38.5714~$3.8571 | 53.5714~$5.3571 | 20% |
Call Alibaba's current Qwen AI models on APIMart: Qwen3.8-Max for complex reasoning and coding, Qwen3.8-Flash for fast, low-cost work, Qwen3.8-27B for open-weight multimodal tasks, and Qwen3.7-Plus and Flash for balanced or high-volume jobs. Use Chat Completions or the Responses API with web search and other built-in tools, and pay per token.
What you get when you call Qwen models through APIMart
How teams use Qwen models in production
Three steps to your first Qwen request on APIMart.
Create an APIMart account and top up your balance. It works for every model on the platform, including the Qwen API.
Generate an API key in your dashboard. The same key calls every Qwen model and all other models on APIMart.
Point the OpenAI SDK at https://api.apimart.ai/v1, set the model to qwen3.8-max and pass stream explicitly, or call /v1/responses for built-in tools.
Call Qwen3.8-Max with the OpenAI-compatible Chat Completions endpoint, or use the Responses API to turn on web search.
Create an APIMart API key, add funds and set APIMART_API_KEY on your server. For Python, install the openai package; both examples use the same key and base URL.
For qwen3.8-max, pass stream explicitly on Chat Completions, because it streams by default when stream is omitted. Built-in tools only work on /v1/responses.
curl --fail-with-body https://api.apimart.ai/v1/chat/completions \
-H "Authorization: Bearer $APIMART_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"model": "qwen3.8-max",
"stream": false,
"messages": [
{
"role": "user",
"content": "Explain when a long-context model is a better fit than retrieval for document Q&A."
}
]
}'import os
from openai import OpenAI
client = OpenAI(
api_key=os.environ["APIMART_API_KEY"],
base_url="https://api.apimart.ai/v1",
)
resp = client.chat.completions.create(
model="qwen3.8-max",
messages=[{"role": "user", "content": "Explain when a long-context model is a better fit than retrieval for document Q&A."}],
stream=False, # qwen3.8-max defaults to streaming when stream is omitted
)
print(resp.choices[0].message.content)import os
from openai import OpenAI
client = OpenAI(
api_key=os.environ["APIMART_API_KEY"],
base_url="https://api.apimart.ai/v1",
)
resp = client.responses.create(
model="qwen3.8-max",
input="Summarize this week's most important AI model releases in three bullet points.",
tools=[{"type": "web_search"}], # built-in tools work only on /v1/responses
max_output_tokens=600,
)
print(resp.output_text)What developers say about our Qwen API
“Qwen-Max is our go-to for Chinese content generation. The language quality is unmatched.”
Wei Zhang
Content Lead
“Qwen-Coder handles our code review pipeline beautifully. Great balance of speed and accuracy.”
Alex Chen
Engineering Manager
“The Qwen-VL model reads Chinese documents and forms with impressive accuracy. Saved us months of development.”
Sarah Li
Product Manager
“Qwen-Turbo is incredibly cost-effective for our high-volume chat application.”
Mike Wang
Backend Developer
“Qwen3 is a huge leap forward. The thinking variants handle complex queries our previous models couldn't.”
Emily Zhou
AI Researcher
“Best Chinese-English translation API we've found. Qwen-MT understands context and idioms remarkably well.”
David Liu
Localization Manager
Choose between the Qwen 3.8 models (Qwen3.8-Max, Qwen3.8-Flash and Qwen3.8-27B) and Qwen3.7-Plus and Qwen3.7-Flash. Compare APIMart token rates, context windows and use cases, then select a model to open its playground and detailed pricing.
Prices are APIMart Gold member rates in USD per 1M tokens. Input and output are billed separately; the cache-read price applies only to cache hits. Built-in tool calls and official prices are listed in the detailed pricing above.
| Model / API ID | Input | Output | Cache read | Context window | When to choose it |
|---|---|---|---|---|---|
Qwen3.8-Maxqwen3.8-max | $1.3714 | $4.1143 | $0.1714 | 983,616Max output 131,072 | Complex reasoning, coding and multi-step agent work. |
Qwen3.8-Flashqwen3.8-flash | $0.0914 | $0.3086 | $0.0114 | 1,000,000Max output 131,072 | Fast, low-cost chat, extraction and pipelines. |
Qwen3.8-27Bqwen3.8-27b | $0.3429 | $1.3714 | $0.0686 | 262,144Max output 131,072 | Image and video understanding with an open-weight model. |
Qwen3.7-Plusqwen3.7-plus | $0.2286 | $0.9146 | — | 1,000,000Max output 131,072 | A balance of quality, speed and cost. |
Qwen3.7-Flashqwen3.7-flash | $0.0229 | $0.0914 | $0.0046 | 1,000,000 | The lowest cost for simple, high-volume tasks. |
APIMart routes Qwen requests to Alibaba's official API, so context windows, max output and model features match Alibaba's published specs; only the price differs. Qwen3.8-27B's limits come from its official model card. A dash means the value is unavailable.
Common questions about the Qwen API on APIMart
APIMart offers Qwen3.8-Max (qwen3.8-max, plus the dated snapshot qwen3.8-max-0902 and the open-weight checkpoint qwen3.8-2.4t-a95b), Qwen3.8-Flash (qwen3.8-flash), Qwen3.8-27B (qwen3.8-27b), Qwen3.7-Max, Qwen3.7-Plus and Qwen3.7-Flash, Qwen3.6-Plus and Qwen3.6-Flash, and qwen-turbo, covering the Qwen3.8, Qwen3.7 and Qwen3.6 generations of the Qwen3 series. Use these IDs as the model parameter; the playground above shows the full Qwen model list.
Start with Qwen3.8-Max for complex reasoning, coding and agent work. Use Qwen3.8-Flash when speed and cost matter, Qwen3.8-27B for image and video understanding, Qwen3.7-Plus for a balance of quality and price, and Qwen3.7-Flash for simple, high-volume tasks.
Choose Qwen3.8 for new projects. Qwen3.8-Max is the newer flagship, and Alibaba lists it below Qwen3.7-Max ($2 / $6 vs $2.5 / $7.5 per million input and output tokens); on APIMart both cost $1.37 / $4.11 for Gold members. Keep Qwen3.7-Plus for a balance of quality and cost, and Qwen3.7-Flash for the lowest price.
Qwen3.8-Max, released by the Qwen team on August 2, 2026, is the most capable Qwen model so far, and Qwen3.8-27B followed in August 2026. Alibaba has said Qwen 4 is still in training and has not released it, so it is not available on APIMart yet.
The Qwen API is billed per million input and output tokens, and each model has its own rates. As of November 2026, Gold member prices are $1.37 input / $4.11 output for Qwen3.8-Max, $0.09 / $0.31 for Qwen3.8-Flash, $0.34 / $1.37 for Qwen3.8-27B, $0.23 / $0.91 for Qwen3.7-Plus and $0.02 / $0.09 for Qwen3.7-Flash. Built-in tools are billed per call. Check the pricing table above for current prices.
Yes. Alibaba Cloud's international list price for Qwen3.8-Max is $2 / $6 per million input and output tokens, and $0.15 / $0.47 for Qwen3.8-Flash. On APIMart, Gold members pay about 31% less for Qwen3.8-Max, Platinum and Diamond members save more, and you only pay for successful requests.
No. APIMart does not offer free credits for the Qwen API; Qwen models are pay-as-you-go by token usage. Alibaba Cloud Model Studio gives new users a free token quota on some models, but that quota only applies to Alibaba's own console.
Sign up for APIMart, add funds, open the API Keys page in your dashboard and create a new key. The same key calls every Qwen model and all other models on the platform, with pay-as-you-go billing and no subscription. An APIMart key is separate from an Alibaba Cloud Model Studio key.
Yes. Sign in to APIMart, create an API key and add funds, then use the playground at the top of this page to chat with Qwen in your browser. Playground messages use your API key and are billed per token from your balance, at the same rates as the API.
Yes. Your APIMart balance works as prepaid credits: top up once, and each Qwen request is deducted per token at the rates in the pricing table. The same credits pay for every other model on APIMart, and there is no monthly plan.
Yes. Point the OpenAI SDK at https://api.apimart.ai/v1, use your APIMart key and set the model to a Qwen ID such as qwen3.8-max. One difference: for qwen3.8-max, Chat Completions streams by default when stream is omitted, so pass stream=False for a normal JSON response.
Use Chat Completions (/v1/chat/completions) for regular conversations and PDF input. Use the Responses API (/v1/responses) for built-in tools such as web search, web extraction, code interpreter and image search; tools sent to Chat Completions are ignored without an error.
Each tool call is billed per call on top of the normal token usage, at the rates shown in the Qwen3.8-Max pricing section. The usage.x_tools field in the response is the billing source of truth. web_extractor must be used together with web_search.
On Qwen3.8-Max, thinking is always on and cannot be disabled, and reasoning tokens count toward output tokens. Qwen3.8-27B has thinking on by default and can turn it off per request, according to the Qwen model card.
Qwen3.8-Max accepts up to 983,616 input tokens and 131,072 output tokens. Qwen3.8-Flash, Qwen3.7-Plus and Qwen3.7-Flash have 1M-token context windows. APIMart routes Qwen requests to Alibaba's official API, so the same limits apply on APIMart.
qwen3.8-max is the main Qwen3.8-Max model ID, and qwen3.8-max-0902 is a dated snapshot of it. qwen3.8-2.4t-a95b is the open-weight checkpoint of the same 2.4T-parameter, 95B-active model. Alibaba lists all three at the same price.
No. Qwen3-Max is an earlier generation of Alibaba's Max-tier model. The current Max tier is Qwen3.8-Max (qwen3.8-max), released on August 2, 2026, and APIMart also offers Qwen3.7-Max. Qwen3-Max itself is not available on APIMart, so use qwen3.8-max for new projects.
Some Qwen models have open weights: Qwen3.8-27B is released under Apache 2.0, and Qwen3.8-2.4T-A95B is on Hugging Face under its own license. Running them yourself needs your own GPUs; the 2.4T model needs data-center multi-GPU hardware. The Qwen API on APIMart gives you the same models without hosting them.
Yes. Qwen is the model family built by Alibaba's Qwen team; its Chinese name is Tongyi Qianwen (通义千问), often shortened to Qianwen. The Qwen API on APIMart calls the same Alibaba models.
Yes. Qwen3.8-Max supports implicit caching of repeated prefixes and explicit caching for content you mark. Cache hits are billed at a lower rate than standard input; see the pricing table above.
Yes. On APIMart, qwen3.8-max accepts images and video as input (for example input_image) and reads PDFs on the Chat Completions endpoint, where PDF pages are billed as image input tokens. If you need open weights, Qwen3.8-27B is a native vision-language model that also understands images and video. Test with the playground above before sending large files.
No. Under APIMart's Terms of Service, your data is not used to train AI models or shared with third parties except as needed to provide the service. Requests are processed only to serve your API calls.
Explore more models in the same category.
Gemini 4 Hydrogen
Gemini 4 Hydrogen is Google’s fast, low-cost Gemini 4 model for everyday coding and agentic tasks.
Gemini 4 Argon
Gemini-4-Argon is an AI assistant that brings together analytical thinking and creative expression. From untangling complex problems to refining writing and code, it offers clear reasoning, natural conversation, and practical suggestions to help turn rough ideas into actionable next steps.
Claude Haiku 5.5
Claude Haiku 5.5 API on APIMart delivers fast, cost-efficient AI, offering responsive conversations, capable coding assistance, and reliable performance for high-volume workflows.
Claude Opus 5
Claude Opus 5 is Anthropic’s next-generation flagship large language model, featuring enhanced capabilities in code development, complex reasoning, knowledge analysis, and agent execution, making it ideal for large-scale project development and professional work scenarios.