Qwen4-Max is the next-generation flagship multimodal model in the Qwen family, combining text generation, image and video understanding, and deep reasoning. It is designed for coding, office productivity, frontend design, and complex agentic tasks. With support for web search, tool calling, and structured outputs, along with compatibility with the OpenAI and Anthropic APIs, it enables developers to build intelligent applications capable of multimodal understanding and task execution.
model parameter): qwen4-max
Inference provider
Alibaba Cloud Model Studio
Model capabilities
- China (Beijing)
- Singapore
- Germany (Frankfurt)
- US (Virginia)
- Japan (Tokyo)
- Hong Kong (China)
Capability | Support | Capability | Support |
|---|---|---|---|
Input modalities | Text, image, video | Output modality | Text |
Streaming | Supported | deep thinking | Supported |
parallel tool calling | Supported | web search | Supported |
prefix completion | Supported | Function Calling | Supported |
Thinking summary and encryption | Supported | OpenAI Chat Completions, Responses, Anthropic Messages, DashScope | Supported |
Structured output (JSON Object, JSON Schema) | Supported | Explicit cache, implicit cache | Supported |
Batch inference | Supported | fine-tuning | Not supported |
model deployment | Not supported |
Model rate limits and context
Parameter | Token | Parameter | Token |
|---|---|---|---|
Maximum input length | 991,808 | Maximum output length | 131,072 |
Maximum input length (thinking mode) | 983,616 | Maximum output length (thinking mode) | 262,144 |
Context length | 1,000,000 | Maximum chain-of-thought length | 262,144 |
Default parameters
Parameter | Default |
|---|---|
temperature | 1.0 in thinking mode; 0.7 in non-thinking mode. |
top_p | 0.95 in thinking mode; 0.8 in non-thinking mode. |
top_k | 20 |
repetition_penalty | 1.0 |
presence_penalty | 0.0 (Thinking); 1.5 (Non-thinking) |
logprobs | false |
enable_thinking | true |
preserve_thinking | true |
reasoning_effort | high |
thinking_budget | 131072 |
max_completion_tokens | 262144 (Thinking); 131072 (Non-thinking) |
reasoning_effort and thinking_budget are omitted, the defaults are reasoning_effort=high and thinking_budget=131072.
Thinking effort
| Protocol | Parameter | Supported values | Default | Mapping |
|---|---|---|---|---|
| OpenAI Chat Completions / DashScope | reasoning_effort | low, high, xhigh | high | minimal → low; medium → high; max → xhigh |
| OpenAI Responses | reasoning.effort | low, high, xhigh | high | minimal → low; medium → high; max → xhigh |
| Anthropic Messages | output_config.effort | low, high, xhigh | high | medium → high; max → xhigh |
Model pricing
- China (Beijing)
- US (Virginia)
- Singapore
- Germany (Frankfurt)
- Japan (Tokyo)
- Hong Kong (China)
| Model ID | Mode | Input tokens per request | Input price (per 1 million tokens) | Output price (per 1 million tokens)Chain of thought + answer | Free quota(Note)Valid for 90 days from the date of Model Studio activation, model release, or application approval, whichever is later |
|---|---|---|---|---|---|
qwen4-maxCache price (per million tokens) | Non-Thinking and Thinking modes | 0<Token≤1M | CNY 10 | CNY 30 | 1 million tokens |
| API | Input | Output |
|---|---|---|
| Batch File | CNY 5 | CNY 15 |
| Batch Chat (limited-time 50% off) | Original price: CNY 10 | Original price: CNY 30 |
Rate limits
- China (Beijing)
- Singapore
- US (Virginia)
- Germany (Frankfurt)
- Japan (Tokyo)
- Hong Kong (China)
Bailian platform monthly consumption: | Tier 1 (current) ≤¥100,000 | Tier 2 ¥100,000~¥1,000,000 | Tier 3 >¥1,000,000 |
|---|---|---|---|
Supported models | TPM rate limit tiers | ||
qwen4-max | 5,000,000 | 10,000,000 | 20,000,000 |
API access
- OpenAI Chat Completions
- OpenAI Responses
- Anthropic Messages
- DashScope
- Model capabilities
- CoT Summary
- Prompt Cache
Usage
Explicit caching
Reuse common input prefixes such as system prompts, tool descriptions, and reference documents to reduce repeated input costs. Supported by OpenAI Chat Completions, Responses, and Anthropic Messages, with TTLs of 5m and 1h. See Explicit caching.