Which is better, Qwen3.6 Plus or Qwen3.7 Plus Preview?

Across 52 shared benchmarks tracked on Vector Wire, Qwen3.6 Plus leads on 17 and Qwen3.7 Plus Preview leads on 33 — Qwen3.7 Plus Preview leads on more. Strengths differ by axis; the table below shows every head-to-head number with provenance.

SEEDRefresh Models —Deals —Regulatory —Sources —

Qwen3.6 Plus vs Qwen3.7 Plus Preview

AlibabavsAlibaba52 shared benchmarks17–33 head-to-head

Benchmark	Qwen3.6 Plus	Qwen3.7 Plus Preview
AA Agentic Index	29	20.8
AA Intelligence	40.5	39.4
AA-LCR	72.3	69
AA-Omniscience	2.6	2.4
arena_elo	1444	1458
arena_text_factuality	1438	1454
Artificial Analysis Coding Index	54.5	55.9
Claw Eval (pass@3)	58.7	62.7
critpt	2.9	9.1
DeepPlanning	41.5	62.3
ERQA	65.7	69.8
Finance Agent v2	40.9	38.2
gdpval	32	22.2
GPQA Diamond	90.4	90.3
HLE	28.8	35.6
hmmt_feb_2026	87.8	92.9
IFBench	75.2	79.1
ifeval	94.3	94.6
imo_answer_bench	83.8	86
include	85.1	83
LiveCodeBench v6	87.1	89.6
mathvision	88	90.3
MCP Atlas	74.1	73.2
mcpmark	48.2	58.7
MLVU	86.7	87.4
mmlu_prox	84.7	85.4
mmlu_redux	94.5	94.5
MMLU-Pro	88.5	88.5
mmmlu	89.5	89
MMMU-Pro	78.8	80.5
nl2repo	37.9	41.1
OmniDocBench 1.5	91.2	91.4
OmniScience Accuracy	26.4	22.5
OmniScience Non-Hallucination	68	74.5
OSWorld-Verified	62.5	73.3
RealWorldQA	85.4	86.9
scicode	40.7	51.3
screenspot_pro_no_tools	68.2	79
SimpleVQA	0.7	0.8
supergpqa	71.6	71.4
SWE-bench Multilingual	73.8	75.8
SWE-bench Pro	56.6	57.6
SWE-bench Verified	78.8	77.7
TauBench V3 - Banking	20.8	17.9
Terminal-Bench 2.0	61.6	70.3
Terminal-Bench 2.1	61.4	61
Terminal-Bench Hard	43.9	47
Video-MME	84.2	88
VideoMMMU	84	85.4
VITA-Bench	44.3	45.6
τ²-Bench Telecom (AA run)	97.7	93
τ³-Bench	70.7	17.9

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.