Qwen4-Max API pricing per million tokens
Qwen4-Max costs $1.50 per million input tokens and $4.50 per million output, with a 1M context window.
AIGROW credit covers Claude Opus 5.5, Claude Fable 5.2, GPT-6.2 Astra, Claude Sonnet 5.5, Claude Haiku 5.5, GPT-6.2 Sol, GPT-6.1 Terra, GPT-6 Luna, Grok 4.7, Gemini 4 Argon, Kimi K3.1 Pro, Gemini 4 Hydrogen, Gemini 3.8 Flash, Claude Fable 5.1 (previous version), GPT-6 Astra (previous version), GPT-6.1 Sol (previous version) and Qwen3.8 Max (previous version). See API credit
$25 a month brings $110 of API credit and cancels any time; $10 once buys $40. US dollars per million tokens, standard rate, shortest context band.
- 1,000 support replies
- $6.30
- Cached input
- $0.075
- Context window
- 1,000,000 tokens
- Released
- 16 Nov 2026
- Weights
- Closed weights
Qwen4-Max costs $1.50 per million input tokens and $4.50 per million output tokens on Alibaba (Qwen)’s own API, and $0.075 for cached input. At those prices 1,000 support replies cost $6.30, the 14th most expensive of 35 priced models on our list.
AIGROW API credit per million input / output tokens: Claude Opus 5.5 $1.60 / $8.00, Claude Fable 5.2 $3.20 / $16.00, GPT-6.2 Astra $2.00 / $10.00, Claude Sonnet 5.5 $0.80 / $4.00, Claude Haiku 5.5 $0.04 / $0.20, GPT-6.2 Sol $0.50 / $2.50, GPT-6.1 Terra $0.25 / $1.25, GPT-6 Luna $0.025 / $0.125, Grok 4.7 $0.50 / $1.50, Gemini 4 Argon $1.00 / $5.00, Kimi K3.1 Pro $1.00 / $5.00, Gemini 4 Hydrogen $0.75 / $3.75, Gemini 3.8 Flash $0.375 / $1.875, Claude Fable 5.1 (previous version) $4.00 / $20.00, GPT-6 Astra (previous version) $2.50 / $12.50, GPT-6.1 Sol (previous version) $0.50 / $2.50 and Qwen3.8 Max (previous version) $1.00 / $3.00. That is up to 75% below list price. See API credit
What a job costs
Three jobs, 1,000 runs of each, at the list price above, and where each puts Qwen4-Max among the 35 current models with a list price.
| Job | 1,000 runs | Among 35 |
|---|---|---|
| Support reply3,000 in, 400 out. | $6.30 | 14th most expensive |
| Document summary25,000 in, 1,000 out. | $42.00 | 14th most expensive |
| Agent task80,000 in (70,000 cached), 3,000 out. | $33.75 | 16th cheapest |
The token counts are ours, chosen to look like real work; the prices are Alibaba (Qwen)’s. All three jobs stay under 100,000 tokens a request, below every long-context surcharge on the list.
The small print
What the base rate above leaves out, copied from the maker’s pages: surcharges, discounts, promotions and cache pricing.
Alibaba calls it the next-generation flagship multimodal model in the Qwen family. International (Singapore) deployment, one tier up to 1M input tokens; China (Beijing) lists $1.413 / $4.24. Cache write $1.875 (5 min) or $3 (1 h); an explicit or implicit cache read costs $0.075. Batch inference is not supported on the International scope. Max output 131,072, or 262,144 in thinking mode. Release date is the International launch on Alibaba's newly released models page.
Qwen4-Max specifications
The limits and features Alibaba (Qwen) documents for Qwen4-Max, read on 18 November 2026.
- Maximum output
- 131,072 tokens
- Maximum input
- 991,808 tokens
- Input types
- Text, images and video
- Reasoning
- On, with adjustable effort. Effort levels: low, high, xhigh
- Function calling
- Yes
- Structured output
- Yes
- Web search
- Yes
- Code execution
- Yes
- Prompt caching
- Automatic and with cache markers. Cache lifetimes: 5 minutes and 1 hour
- Batch discount
- 50% off
- Status
- Current
- Fine-tuning
- Not offered on Alibaba (Qwen)’s API
Every fact above was read on these pages, and a fact they do not state is left out:alibabacloud.com/help/en/model-studio/qwen4-maxalibabacloud.com/help/en/model-studio/newly-released-modelsalibabacloud.com/help/en/model-studio/qwen-api-via-openai-responsesalibabacloud.com/help/en/model-studio/explicit-cache-qwen4
Qwen4-Max against the models it is weighed with
Each comparison puts both list prices side by side and costs the same three jobs on each.
- vs Qwen3.8 Max
- Alibaba's Qwen4-Max against the Qwen3.8 Max it replaces, a quarter cheaper on input and on output.
- vs DeepSeek V4 Pro
- Two Chinese frontier APIs, with Qwen4-Max 14% above DeepSeek's peak rate and over twice its off-peak one.
- vs Kimi K3.1 Pro
- Two Chinese flagships, with Qwen4-Max at three quarters of the Kimi input price and under half its output price.
- vs Claude Opus 5.5
- Alibaba's flagship Qwen4-Max against Anthropic's Opus 5.5, at 37.5% of the Opus input price and 22.5% of its output price.
- vs Gemini 4 Hydrogen
- Alibaba's flagship Qwen4-Max and Google's Gemini 4 Hydrogen at the same $1.50 input and $0.075 cache read, with Qwen 40% cheaper on output.
- vs Claude Fable 5.2
- Alibaba's flagship Qwen4-Max against Anthropic's Fable 5.2, at 18.75% of the Fable input price and 11.25% of its output.
- vs GPT-6.2 Astra
- Alibaba's flagship Qwen4-Max against OpenAI's GPT-6.2 Astra, at 18.75% of the Astra input price and 11.25% of its output.
Other Alibaba (Qwen) models
| Model | Input | Output | Context | 1,000 replies |
|---|---|---|---|---|
| Qwen3.8 FlashAlibaba (Qwen) | $0.15 | $0.47 | 1M | $0.64 |
| Qwen3.7 PlusAlibaba (Qwen) | $0.40 | $1.60 | 1M | $1.84 |
| Qwen3.8 27BAlibaba (Qwen) · open weights | $0.50 | $3.00 | 1M | $2.70 |
| Qwen3.8 MaxAlibaba (Qwen)Replaced by Qwen4-Max | $2.00 | $6.00 | 1M | $8.40 |
Where the price comes from
- Alibaba (Qwen)
- alibabacloud.com/help/en/model-studio/model-pricing
Read on 18 November 2026. Each price is the standard rate for the shortest context band, with no batch or priority discount; long-context surcharges, batch rates and promotions are in each model’s small print. No price was taken from a reseller.
Compare Qwen4-Max with 63 other models
The price list sets every model’s input and output price side by side, with what a support reply, a document summary and an agent task cost on each.
Prices checked 18 November 2026 · See them by release date