Last verified
QWEN MAX FLAGSHIP1M CONTEXTTEXT + VISION + VIDEOITEMIZED CACHETHINKING SAME PRICE

Qwen4 Max API Pricing

Qwen4 Max is Alibaba's new Qwen Max model, released on the International list on November 16, 2026. It lists $1.5/M input and $4.5/M output for 0-1M tokens, published directly in US dollars. That is below Qwen3.8 Max at $2/$6, which stays on the list. Unlike Qwen3.8 Max, it has an itemized cache rate: a hit bills at $0.075/M. Thinking and non-thinking modes bill at the same rate.

Input - per 1M tokens
$1.50/M
Base International row standard
Output - per 1M tokens
$4.50/M
Output incl. thinking tokens standard
Cached input - per 1M tokens
$0.075/M
Hit explicit or implicit, creation 125% 95% off
Effective - agentic blend
$0.727/M
92/8 split - 82% cache
§ 01 / TERMINAL

Qwen4 Max API cost calculator.

Calculator pre-loaded with the International Qwen4 Max rates. A cache hit bills at $0.075/M and the calculator counts every miss at the 5-minute creation rate, $1.88/M, so the figure is the cost of explicit caching; implicit caching has no creation charge and comes out lower.

$ /mo
Workload split
Prompt cache hit rate
Tokens you can process
—
Words equivalent (English)
—
Effective rate
—
Open full calculator (all models · share URL · CSV) →
§ 02 / SCENARIOS

Qwen4 Max cost on real workloads.

§ 03 / TOKENIZER

Qwen4 Max token counter and cost.

What's a token →
Calibrated · measured on the vendor's tokenizer · 2026-06-10 Auto-counts as you type

Counts use a chars-per-token calibration measured on the vendor's own published tokenizer (Qwen/Qwen3.5-397B-A17B, 2026-06-10). English prose is typically within a few percent; code and non-Latin scripts tokenize heavier. For billing-exact counts use the vendor's count-tokens API.

Characters 363
Words 58
Tokens (estimated) 69 tokens
Cost as input · uncached $0.0001 USD
Cost as output · uncached $0.00031 USD
Cost as cached input $0.00001 USD
§ 04 / SHELF

Qwen4 Max vs similar models.

All models →
Model Input /M Output /M Effective blended Context Best for
Qwen4 Max Current $1.50 cache $0.075 $4.50 $0.727 agentic 92/8 1M Newest Qwen Max, with an itemized cache rate
Qwen3.8 Max $2.00 $6.00 $2.32 +3.2× 1M Previous Max, higher list rate, cache not itemized
Qwen3.7 Max $2.50 $7.50 $2.90 +4.0× 1M Older Max, text only
DeepSeek V4 Pro $1.32 cache $0.044 $3.96 $0.569 −22% 1M Cheaper Chinese frontier tier, text only
GPT-6.2 Sol $2.00 cache $0.10 $10.00 $1.29 +77% 1.05M OpenAI's balanced tier for coding and agents
Gemini 4 Carbon $2.00 cache $0.10 $10.00 $1.21 +66% · promo 1M Google's Gemini 4 with a real cache discount
§ 05 / DEEP LINKS

Qwen4 Max cost by scenario.

All calculators →

Qwen4 Max API pricing FAQ.

Short answers for teams checking Qwen4 Max pricing, its cache rates, its limits and how it compares with the Max it follows.

Q · 01 What is Qwen4 Max priced at? +
Alibaba's International (Singapore) rate card lists $1.5/M input and $4.5/M output, published directly in US dollars, in a single billing band covering 0-1M input tokens per request. Under AI//COST's 92/8 agentic blend with an 82% cache hit rate the effective planning figure is $0.727/M.
Q · 02 How much does context caching cost? +
Alibaba publishes a cache table for Qwen4 Max on the International card. A cache hit, explicit or implicit, bills at $0.075/M, 5% of standard input. Creating an explicit cache block bills at $1.88/M for a 5-minute block and $3/M for a 1-hour block. Implicit caching has no creation charge. Qwen3.8 Max and Qwen3.7 Max carry only a note that caching is discounted, with no rate.
Q · 03 Is it cheaper than Qwen3.8 Max? +
Yes, on both sides: $1.5 against $2 input and $4.5 against $6 output. It also has the itemized cache rate that Qwen3.8 Max lacks. Qwen3.8 Max stays on Alibaba's list; nothing was retired with this release.
Q · 04 Why do I see $1.413 and $4.24 elsewhere? +
Because Alibaba publishes several rate cards on one page and they are different products. $1.413/$4.24 is the Global deployment card; $1.5/$4.5 is the International (Singapore) card. AI//COST quotes the International list for every Alibaba model, so the numbers stay comparable across our catalogue.
Q · 05 Is there a Batch discount? +
Not one we can quote. Alibaba's model page says Qwen4 Max supports batch inference, but the International rate card publishes no batch row for it, and the batch figures elsewhere on the page belong to another deployment scope. We record no batch rate rather than assume a discount.
Q · 06 Does thinking mode cost extra? +
No. Thinking and non-thinking modes share one rate, and the reasoning tokens bill as output at $4.5/M. The reasoning_effort parameter takes low, high or xhigh, with high as the default, so a higher setting costs more through longer output rather than a higher rate.
Q · 07 What are the context and output limits? +
The context window is 1,000,000 tokens. Maximum input is 991,808 tokens, or 983,616 in thinking mode. Maximum output is 131,072 tokens, or 262,144 in thinking mode.
Q · 08 Which API model ID should I use? +
qwen4-max. A qwen4-max-preview ID appeared on November 11, 2026, but only through Alibaba's Token Plan subscription, with no per-token price. Alibaba lists the model as compatible with the OpenAI Chat Completions and Responses APIs and the Anthropic Messages API.
Q · 09 Is there a free quota? +
Yes. The International card lists a free quota of 1 million tokens, valid for 90 days from Model Studio activation or the model's release, whichever is later.