Chat Completions
/v1/chat/completionsOpenAI-compatible messages interface for direct conversations and multimodal input.
- Standard chat and streaming reasoning content
- PDF understanding
- Image and video input
Input
qwen3.8-max
OpenAI-compatible messages interface for direct conversations and multimodal input.
Output
What's on your mind?
Ask 光谱汇深
qwen3.8-max · /v1
Qwen3.8-Max integration and pricing
Choose the endpoint for your workload, then review token, cache, and built-in tool charges in one place.
Choose the right endpoint
Route each workload by capability: conversations and PDF input use Chat Completions; built-in tools use Responses.
Responses API
/v1/responsesUnified input and output interface required for Qwen3.8-Max built-in tools.
- Web search and web extraction
- Code execution
- Web image search by text and reverse image search
Chat Completions defaults to streaming when stream is omitted. Set stream: false explicitly when your client expects one complete JSON response.
PDF understanding is supported only by Chat Completions. The Responses API silently ignores PDF input without returning an error.
web_extractor must be declared together with web_search. Unsupported tool names may be silently accepted but will never run.
Declaring a tool does not incur a charge. Billing uses only actual calls reported in usage.x_tools, independently of token charges.
qwen3.8-max
Qwen3.8-Max pricing
Token prices are per 1 million tokens. Built-in tools are billed by their own usage unit.
Savings 20%
- Standard input
- 13.7143Creditsper 1M tokens
~¥1.3714
- Implicit cache hit
- 1.7143Creditsper 1M tokens
~¥0.1714
- Explicit cache hit
- 1.1429Creditsper 1M tokens
~¥0.1143
- Cache write
- 17.1429Creditsper 1M tokens
~¥1.7143
- Output and reasoning
- 41.1429Creditsper 1M tokens
~¥4.1143
Input tokens that do not hit a cache.
List price: 17.1429 Credits−20%
Matching prompt prefixes are cached automatically; hit tokens are rounded in blocks and may not cover the entire input.
List price: 2.1429 Credits−20%
A reusable ephemeral cache you mark yourself, with a lower hit price than implicit caching.
List price: 1.4286 Credits−20%
Charged when an explicit cache is created; cached content must contain at least 1,024 tokens and is retained for 5 minutes.
List price: 21.4286 Credits−20%
Reasoning cannot be disabled and its tokens are billed as output. A missing output_thinking rate means the single output rate applies, not that reasoning is free.
List price: 51.4286 Credits−20%
Public prices use the default group. Your final charge follows your account group and may be lower than the displayed price.
Context and cache limits
- Maximum input tokens
- 983,616
- Maximum output tokens
- 131,072
- Cache read support
- Supported
- Cache write support
- Supported
Built-in tools
web_search- Your price
- 4.5714Creditsper 1,000 calls
~¥0.4571
- List price
- 5.7143 Credits
web_extractor- Your price
- Free
0 Credits · ~¥0
- List price
- —
code_interpreter- Your price
- Free
0 Credits · ~¥0
- List price
- —
web_search_image- Your price
- 27.4286Creditsper 1,000 calls
~¥2.7429
- List price
- 34.2857 Credits
image_search- Your price
- 54.8571Creditsper 1,000 calls
~¥5.4857
- List price
- 68.5714 Credits
| Built-in tools | Your price | List price |
|---|---|---|
Web searchweb_search | 4.5714Creditsper 1,000 calls ~¥0.4571 | 5.7143 Credits |
Web extractionweb_extractor | Free 0 Credits · ~¥0 | — |
Code Interpretercode_interpreter | Free 0 Credits · ~¥0 | — |
Web image search by textweb_search_image | 27.4286Creditsper 1,000 calls ~¥2.7429 | 34.2857 Credits |
Reverse image searchimage_search | 54.8571Creditsper 1,000 calls ~¥5.4857 | 68.5714 Credits |
Qwen 3.8 Max API and Qwen Model Family
Build with qwen3.8-max for long-context reasoning. Use Chat Completions for conversations and PDFs, or the Responses API for web, extraction, code, and image-search tools.
50K+
Active Users
99.9%
Uptime
2x
Faster
70%
Cost Savings
Qwen 3.8 Max API Capabilities
Endpoint behavior, context limits, reasoning, caching, and tools you should know before integrating
983,616-Token Input
Process up to 983,616 input tokens and generate up to 131,072 output tokens with qwen3.8-max.
Chat and PDF Inputs
Use Chat Completions for standard conversations and PDF inputs. PDF input is not supported through the Responses API.
Responses API Tools
Use web_search, web_extractor, code_interpreter, web_search_image, and image_search through the Responses API.
Required Reasoning
qwen3.8-max always performs reasoning. Reasoning tokens are part of output-token usage and billing.
Explicit Streaming Control
The interface works with the OpenAI SDK, but Chat Completions has a different streaming default, so set stream explicitly.
Implicit and Explicit Cache
The model supports both implicit and explicit context caching. Applicable cache charges appear in the live pricing section.
Web Extraction Dependency
web_extractor depends on web_search. Include web_search when building a Responses API extraction workflow.
Qwen Family Options
Use qwen3.8-max for the capabilities described here, or compare other Qwen models whose limits and supported features may differ.
What to Build with Qwen 3.8 Max
Choose the endpoint that matches each long-context, document, or tool-assisted workflow
Long-Document Analysis
Review large reports, knowledge bases, or multi-file context within the 983,616-token input limit.
Code Investigation
Use code_interpreter through the Responses API for tool-assisted analysis and computational workflows.
Conversational Assistants
Build multi-turn assistants with Chat Completions and explicitly choose whether responses should stream.
PDF Understanding
Send PDF content through Chat Completions for document question answering and extraction.
Web Research
Combine web_search and web_extractor through the Responses API to find and process current web content.
Image Search Workflows
Use web_search_image or image_search through the Responses API when a task needs image discovery.
Get Started with the Qwen API
Choose the model and endpoint deliberately before sending your first request
1
Create an Account and Add Funds
Create an 光谱汇深 account and add balance for API usage.
2
Get an API Key
Generate an API key for authenticated Qwen requests.
3
Choose the Endpoint
Set the model to qwen3.8-max. Use Chat Completions for chat or PDF, Responses for built-in tools, and set stream explicitly for Chat requests.
User Reviews
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
Frequently Asked Questions
Integration details for qwen3.8-max and the Qwen model family
What does this Qwen page focus on?
It is a Qwen family page, with qwen3.8-max as the featured model. Other Qwen models may have different limits, endpoints, and capabilities.
Should I use Chat Completions or the Responses API?
Use Chat Completions for regular conversations and PDF inputs. Use the Responses API for built-in tools such as web search, web extraction, code interpretation, and image search.
Is it compatible with the OpenAI SDK?
The interface supports OpenAI SDK integration, but you should not assume every default is identical. For Chat Completions, pass stream explicitly because its default behavior differs.
How does reasoning affect usage?
Reasoning is required for qwen3.8-max and cannot be disabled. Reasoning tokens count toward output-token usage and billing.
What are the context and output limits?
qwen3.8-max accepts up to 983,616 input tokens and supports up to 131,072 output tokens.
How are pricing and caching shown?
The pricing section loads current model pricing dynamically, including applicable input, output, cache, and tool charges. The model supports implicit and explicit caching.
Is there a discount on the Qwen API?
Yes. The Qwen API is billed per million tokens. The pricing table on this page lists every specification and shows the Gold member price, 20% off. Platinum and Diamond members save even more, up to 28%. You only pay for successful requests.
How do I get a Qwen API key?
Sign up for 光谱汇深, open the API Keys page in your dashboard and create a key. The same key calls the Alibaba Qwen API and every other model on the platform, with pay-as-you-go billing and no subscription.
How do I use the Qwen 3.8 Max API?
Call the OpenAI-compatible Chat Completions endpoint with your 光谱汇深 API key and the model set to qwen3.8-max. For built-in tools such as web search, use the Responses API instead. Full parameters and code examples are in the 光谱汇深 docs (docs.callhub.cn).
How can I contact support for help?
You can submit a support ticket from the dashboard or email us at support@apimart.ai. Our team will get back to you as soon as possible.
Related Models
Explore more models in the same category.
Kimi K3
Kimi-K3 is a next-generation large language model launched by Moonshot AI. It features ultra-long context, multimodal understanding, and strong coding capabilities, making it suitable for complex reasoning, software development, knowledge analysis, and agent automation tasks.
Save 20%IN24Credits/M~¥2.4/MOUT120Credits/M~¥12/M- X
Mimo V2.5 Pro
Xiaomi has launched a high-performance large language model that enhances reasoning, coding, and multitasking capabilities, making it suitable for complex analysis, programming, and agent-based scenarios.
Qwen 3.7 Max
Launched by Alibaba Cloud Tongyi Qianwen, this flagship large language model features enhanced capabilities in reasoning, coding, and handling complex tasks, making it suitable for enterprise-level applications, chatbots, and high-difficulty task scenarios.
Kimi K2.6
A general-purpose large language model launched by Moonshot AI, featuring strong reasoning, coding, and agent capabilities, making it suitable for handling complex tasks.