Which is better, Qwen3.5 27B or Qwen3.6 Plus?

Across 57 shared benchmarks tracked on Vector Wire, Qwen3.5 27B leads on 7 and Qwen3.6 Plus leads on 49 — Qwen3.6 Plus 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 27B vs Qwen3.6 Plus

AlibabavsAlibaba57 shared benchmarks7–49 head-to-head

Benchmark	Qwen3.5 27B	Qwen3.6 Plus
AA Agentic Index	54.6	29
AA Intelligence	34.6	40.5
AA-LCR	72.3	72.3
AA-Omniscience	-44	2.6
AI2D	92.9	94.4
aime_2026	92.6	95.3
Artificial Analysis Coding Index	34.9	54.5
C-Eval	90.5	93.3
Claw Eval (pass@3)	46.2	58.7
coding_arena_elo	1357	1460
critpt	0.9	2.9
DeepPlanning	22.6	41.5
ERQA	60.5	65.7
gdpval	33	32
GPQA Diamond	85.8	90.4
HLE	48.5	28.8
HMMT 2025	89.8	94.6
hmmt_feb_2026	84.3	87.8
hmmt_nov_2025	89.8	94.6
IFBench	76.5	75.2
ifeval	95	94.3
imo_answer_bench	79.9	83.8
include	81.6	85.1
LiveCodeBench v6	80.7	87.1
longbench_v2	60.6	62
mathvision	86	88
MCP Atlas	68.4	74.1
mcpmark	36.3	48.2
MLVU	85.9	86.7
mmlu_prox	82.2	84.7
mmlu_redux	93.2	94.5
MMLU-Pro	86.1	88.5
mmmlu	85.9	89.5
MMMU	82.3	86
MMMU-Pro	75	78.8
MMStar	81	83.3
nl2repo	27.3	37.9
OmniDocBench 1.5	88.9	91.2
OmniScience Accuracy	20.7	26.4
OmniScience Non-Hallucination	24.9	68
OSWorld-Verified	56.2	62.5
RealWorldQA	83.7	85.4
scicode	39.5	40.7
screenspot_pro_no_tools	70.3	68.2
SimpleVQA	56	0.7
supergpqa	65.6	71.6
SWE-bench Multilingual	69.3	73.8
SWE-bench Pro	51.2	56.6
SWE-bench Verified	75	78.8
Terminal-Bench 2.0	41.6	61.6
Terminal-Bench Hard	32.6	43.9
tool_decathlon	31.5	39.8
VideoMMMU	82.3	84
VITA-Bench	41.9	44.3
WideSearch	66.4	74.3
τ²-Bench Telecom (AA run)	93.9	97.7
τ³-Bench	68.4	70.7

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.