Which is better, Qwen3.5 35B A3B or Qwen3.7 Plus Preview?

Across 54 shared benchmarks tracked on Vector Wire, Qwen3.5 35B A3B leads on 4 and Qwen3.7 Plus Preview leads on 50 — 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 —

Qwen3.5 35B A3B vs Qwen3.7 Plus Preview

AlibabavsAlibaba54 shared benchmarks4–50 head-to-head

Benchmark	Qwen3.5 35B A3B	Qwen3.7 Plus Preview
AA Agentic Index	44.1	20.8
AA Intelligence	29.9	39.4
AA-LCR	68.3	69
AA-Omniscience	-48.1	2.4
Artificial Analysis Coding Index	37	55.9
baby_vision	38.4	70.4
bfcl_v4	67.3	72.9
charxiv_rq	77.5	85.9
Claw Eval (pass@3)	51	62.7
critpt	0.9	9.1
DeepPlanning	22.8	62.3
ERQA	64.8	69.8
gdpval	20.3	22.2
GPQA Diamond	84.5	90.3
HLE	47.4	35.6
hmmt_feb_2026	81.8	92.9
IFBench	72.5	79.1
ifeval	91.9	94.6
imo_answer_bench	76.8	86
include	79.7	83
LiveCodeBench v6	74.6	89.6
LVBench	71.4	76.2
mathvision	83.9	90.3
MCP Atlas	62.4	73.2
mcpmark	27	58.7
MLVU	85.6	87.4
mmlu_prox	81	85.4
mmlu_redux	93.3	94.5
MMLU-Pro	85.3	88.5
mmmlu	85.2	89
MMMU-Pro	75.1	80.5
nl2repo	20.5	41.1
ocrbench_v2	65.3	67.1
OmniDocBench 1.5	89.3	91.4
OmniScience Accuracy	20.1	22.5
OmniScience Non-Hallucination	14.6	74.5
OSWorld-Verified	54.5	73.3
QwenClawBench	47.7	61.8
RealWorldQA	84.1	86.9
scicode	37.7	51.3
screenspot_pro_no_tools	68.6	79
SimpleVQA	58.3	0.8
supergpqa	63.4	71.4
SWE-bench Multilingual	60.3	75.8
SWE-bench Pro	44.6	57.6
SWE-bench Verified	70	77.7
TauBench V3 - Banking	4.9	17.9
Terminal-Bench 2.0	40.5	70.3
Terminal-Bench 2.1	40.8	61
Terminal-Bench Hard	26.5	47
VideoMMMU	80.4	85.4
VITA-Bench	31.9	45.6
τ²-Bench Telecom (AA run)	89.2	93
τ³-Bench	68.9	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.