Qwen4-Max by Alibaba added to BenchGecko
Alibaba's Qwen4 flagship goes pay per token · $1.50 input and $4.50 output per million tokens · 1M token context · 25% below Qwen3.8 Max list price.
What was added
Qwen4-Max by Alibaba is now listed on BenchGecko. Alibaba released it on November 16, 2026 as the most capable model in the Qwen family so far, five days after the Token Plan only Qwen4-Max Preview. The API model id is qwen4-max, and unlike the preview it is billed per token.
Size and context
Alibaba describes Qwen4-Max as a sparse Mixture of Experts model with 2.4 trillion total parameters and 71 billion activated per token, which it says is 25% fewer activated parameters than Qwen3.8-Max at the same scale. The QwenCloud model page lists a 1M token context window, a maximum input of 991K tokens (983K with thinking), a maximum output of 393K tokens and up to 262K reasoning tokens. It accepts text, images and video and returns text.
Pricing
On QwenCloud International (Singapore), Qwen4-Max costs $1.50 per million input tokens and $4.50 per million output tokens, with thinking tokens billed as output. Implicit and explicit cache reads cost $0.075 per million tokens, and explicit cache creation costs $1.875 (5 minute) or $3 (1 hour) per million tokens. Thinking and non thinking modes share one price with no input length tiers. The Beijing and Global deployment lists $1.413 input and $4.24 output.
For comparison, Qwen3.8 Max and Qwen3.8 Max (0902) are listed on BenchGecko at $2 input and $6 output, so Qwen4-Max is 25% cheaper on both sides at list price. Qwen3.8 Max Prime sits at $4 and $12.
Benchmarks
Alibaba reports that Qwen4-Max ranked 6th on Code Arena and, on its own Work360 benchmark across 721 fields, beat Qwen3.8-Max while using 35% fewer tokens at High reasoning effort and cutting average cost per task by around 60%. These are vendor figures. BenchGecko has no independent scores yet and will add them to the model page as they arrive. Alibaba also says smaller open weight Qwen4 models will follow.
Source: BenchGecko
Written automatically from BenchGecko's tracked data for this period; every number comes from the changelog, model, mindshare and pricing datasets. Checked against the source data before publishing.