Which is better, Qwen3.7 Max or Qwen3.8 Max Preview?

Refresh Models —Deals —Regulatory —Sources —

Qwen3.7 Max vs Qwen3.8 Max Preview

40 SHARED BENCHMARKS

Across 40 shared benchmarks, Qwen3.7 Max scores higher on 8 and Qwen3.8 Max Preview on 32. The widest gap is TauBench V3 - Banking, where Qwen3.8 Max Preview scores 51.3 against 11.8. Qwen3.8 Max Preview is the cheaper of the two on tracked API pricing ($2.00 against $2.50 per million input tokens).

ALIBABAVSALIBABA40 SHARED8–32 HEAD-TO-HEAD

BenchmarkQwen3.7 MaxMARGINQwen3.8 Max Preview

AA Agentic Index30.958.4

AA Intelligence46.758.1

AA-LCR74.774.3

AA-Omniscience13.53.4

Agents' Last Exam (Pass / Score)31.152.4

AndroidBench56.575.1

arena_elo14741481

Artificial Analysis Coding Index6671.8

coding_arena_elo15171691

critpt13.420

DeepSWE1856.6

DeepSWE 1.121.656.6

gdpval38.261

GPQA Diamond92.492.7

HealthBench54.560.2

HLE41.443.6

HLE (with tools)53.556.2

IFBench80.582.8

JobBench31.353.4

livebench_agentic_coding43.664.7

livebench_coding74.272.9

livebench_data_analysis71.878.4

livebench_instruction_following7474.1

livebench_language79.779.7

livebench_math85.391.3

livebench_reasoning83.388.2

longbench_v265.366.3

MLS Bench Lite↓ lower is better31.741

MRCR v2 256K (8-needle)86.792.9

nl2repo47.255.9

OmniScience Accuracy31.131.9

OmniScience Non-Hallucination74.458.3

PaperBench64.893

scicode53.552.9

simpleqa_verified55.845.8

SWE-bench Pro60.667.7

TauBench V3 - Banking11.851.3

Terminal-Bench 2.17586.6

Toolathlon Verified49.772.5

WideSearch75.281.9

Best tracked score per model per benchmark (default configuration; source-attributed). ↓ marks lower-is-better metrics. Open either model for its full surface, provenance and pricing. Quoted rates are the price-setter row we currently track for each model — its direct or vendor-official listing where one exists (alibaba-official, alibaba-official), otherwise the lowest tracked offer.