Tool support requires the appropriate API integration; a supported tool is not automatically active in every chat.
Supported API features and tools
262,144-token text context
Hybrid thinking with non-thinking as the documented default
Structured output and context caching on the Singapore reference deployment
Text-based programming and evidence synthesis workflows
Before you choose
Singapore function calling is listed as unsupported, even though the Beijing table supports it. General agent marketing is not a regional API guarantee.
This is not Qwen3.7-Max and does not inherit its 1M context or output ceiling.
Maximum input is 258,048; answer output is lower in thinking mode. No image or video input is claimed.
The reviewed sources do not establish a knowledge cutoff. Supply current evidence for time-sensitive questions; a large context window is not live web access.
Extended thinking
Thinking disabled · defaultThinking enabled
This model supports thinking and non-thinking requests. Alibaba Cloud's enable_thinking switch controls the mode; in the OpenAI Python SDK it belongs in extra_body. The provider's thinking_budget controls reasoning length. These are integration settings, not features a prompt can activate in every app.
The overview maps this alias to qwen3-max-2026-01-23. Check the resolved version rather than treating the alias as immutable.
The 81,920-token chain-of-thought ceiling is separate from the 32,768 thinking-mode answer limit.
Put it to work
Start with a more useful prompt.
Original examples from EZ Ai Assist. Adapt these to your task and the features available in your workspace.
Workflow 01
Build a source-backed decision memo
Separate recommendation from evidence gaps.
Using only these proposals and requirements, write a decision memo comparing feasibility, operating effort and risks. Cite each substantive claim to its source section, flag missing evidence and distinguish assumptions from conclusions. Finish with one recommended validation step before commitment.
Workflow 02
Review a text-only API contract
Check compatibility before implementation.
Compare the old and new API contracts below. Identify breaking changes, ambiguous validation rules and missing error cases. Cite the relevant fields, propose minimal compatibility tests and explain expected outcomes. Do not assume an implementation exists beyond the supplied contracts.
Workflow 03
Compare reasoning-mode outputs
Evaluate quality without guessing settings.
Score these two model responses against the supplied rubric for correctness, evidence and instruction following. Cite specific passages for each score and identify failures that need human review. Do not infer which response used thinking mode, and leave latency and cost blank unless measured data is supplied.
Developer reference
Alibaba Qwen API pricing
These are provider API reference prices, not EZ Ai Assist subscription prices.
Singapore / International API reference · USD per 1,000,000 tokens
Token type
Price
Input · input ≤32K
$1.20
Output · input ≤32K
$6.00
Input · 32K < input ≤128K
$2.40
Output · 32K < input ≤128K
$12.00
Input · 128K < input ≤256K
$3.00
Output · 128K < input ≤256K
$15.00
Rates describe the Singapore / International deployment, in USD per million tokens, excluding promotions. Other regions, snapshots and hosting services can have different rates and supported features.
Input-length bands refer to the provider's published prompt-size tiers, not the length of the answer. Check the selected region, resolved model version, cache policy and any additional service charges before estimating cost.
For the same three bands, implicit cached input is $0.24 / $0.48 / $0.60; explicit cache creation is $1.50 / $3.00 / $3.75 and reads are $0.12 / $0.24 / $0.30 per million tokens.
Common questions
A few things worth knowing.
Is thinking enabled by default?
The reviewed Alibaba Cloud thinking guide says it is disabled by default for qwen3-max. Set the mode explicitly in controlled evaluations.
Why are two output limits shown?
The model table lists 65,536 tokens without thinking and 32,768 with thinking. Presenting only the larger number would hide an important mode-specific limit.
Can I assume function calling works in Singapore?
No. The current regional table lists it as unsupported there, despite support in Beijing. Verify the selected deployment contract.
Does this page describe the newer 3.7 Max?
No. It describes qwen3-max with 262,144 context; Qwen3.7-Max has its own guide and specifications.
Does this page guarantee access in EZ Ai Assist?
No. A catalog entry and provider capability reference do not establish current app access, enabled tools or available settings. Check the model selector and controls in your workspace.
Are these EZ Ai Assist subscription prices?
No. Provider API rates are developer references. Your subscription and any limits shown in the app are separate; use the website pricing page for plans.
How should I verify an answer?
Provide the relevant evidence and acceptance criteria, ask for source references or reproducible checks, and review the result. Distinguish a proposed tool action from a tool action actually executed.
Check the source
Official documentation
Specifications and API prices checked on . Example prompts and workflow advice are editorial guidance from EZ Ai Assist.