Qwen 4 Max

Qwen 4 Max is a model from Alibaba, released November 15, 2026. It ranks #11 of 38 models on the Vals Index with 63.20%. Its best result is #15 of 129 on Vibe Code Bench v1.1.

Release Date: Nov 15, 2026

Developer Alibaba  🇨🇳
Context Window 1M
Max Output Tokens 262k
Token Costs (in/out) $1.50/4.50
Weights Private
Input Modalities

Accuracy

63.20 % ± 1.01

Cost / Test (Vals Index)

$ 9.330

Latency

64 min 39 s

Vals Index
BenchmarksAccuracyRankings

63.20%

±1.01
11/38

61.99%

±1.45
8/41

70.71%

±1.75
17/39

70.83%

±2.36
18/43

66.68%

±2.02
8/41

41.35%

±3.42
33/95

40.35%

±3.54
13/86

86.66%

±2.27
15/129

21.25%

±3.33
7/39

48.48%

±0.88
15/65
Proprietary BenchmarksAcademic BenchmarksIndustry Partners
Vals
Default Provider : Alibaba
Temperature: 1
Top P: 0.95
Top K: 20
Max Output Tokens: 262,144
Reasoning Effort: xhigh

Updates

Nov 15, 2026

We evaluated Alibaba’s new Qwen 4 Max on Vals Index v2.2, its components and the Vals Finance Index.

Qwen 4 Max is a reasoning model with a 1M-token context window and 262k max output tokens. We evaluated it at xhigh reasoning effort, temperature 1.0, top_p 0.95 and top_k 20.

Congrats to the Qwen team on the release!