Alibaba

Qwen4 Max

Models.dev
Intelligence
42.9
Rank 61 of 813 measured models
Output speed
65.1 t/s
Rank 328 of 476 measured models
Context window
984K
Rank 421 of 1420 measured models
Cost per task
$3.27
Rank 211 of 255 measured models

Qwen4 Max is a text AI model from Alibaba, released on Nov 16, 2026, with a context window of 984K tokens. Qwen4 Max scores 42.9 on the Artificial Analysis Intelligence Index, ranking 61st of 813 measured models. The closest models on this index are Claude Opus 5 (44.2), Muse Wonder (43.6), and Claude Opus 4.8 (batch) (41.8). With an output speed of 65.1 t/s, it is slower than Claude Opus 5 (79.1 t/s) and Muse Wonder (110 t/s). Every benchmark side by side: Qwen4 Max vs Muse Wonder.

Artificial Analysis Indices

Intelligence42.9
Rank 61 of 813 measured models

Capability indexes

Performance in specific industries, measured by Artificial Analysis
Finance & Accounting54.1
Rank 33 of 283 measured models
Strategy & Ops56.9
Rank 32 of 283 measured models
Legal49.6
Rank 49 of 283 measured models
Healthcare & Medical48.2
Rank 36 of 210 measured models
Engineering50.1
Rank 51 of 281 measured models
Economics51.3
Rank 51 of 289 measured models

Standard Benchmarks

FrontierCode 2.026.5%
Rank 20 of 37 measured models
Terminal-Bench-Science23.3%
Rank 16 of 37 measured models
HLE-Diamond44.9%
Rank 14 of 37 measured models
CritPt v2 (Hard Subset)43.0%
Rank 18 of 33 measured models
AutomationBench-AA72.6%
Rank 7 of 37 measured models
GDP.pdf (all-pass)26.2%
Rank 12 of 37 measured models
Terminal-Bench 4.048.5%
Rank 31 of 250 measured models
AA-Omniscience43.7%
Rank 29 of 74 measured models

Models at a similar level

The closest models on the Intelligence Index, one configuration per family.

Preferences and analytics

Rybbit: visits, performance, clicks, copies, forms and session replay (masked inputs), with your permission. Refusing does not limit the site.

Privacy

Allow navigation analytics?

With your permission, Rybbit measures visits, performance and interactions (clicks, copies, forms), including session replay, to improve BenchSift. Input values are masked in replays. Refusing does not limit the site.

Your choice is stored for 180 days. Change it at any time using “Privacy preferences” in the footer. Withdrawal reloads the page.

No choice stored: Rybbit is disabled.

Privacy