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

Across 52 shared benchmarks tracked on Vector Wire, Qwen3.6 27B leads on 6 and Qwen3.6 Plus leads on 46 — 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 —

Qwen3.6 27B vs Qwen3.6 Plus

AlibabavsAlibaba52 shared benchmarks6–46 head-to-head

Benchmark	Qwen3.6 27B	Qwen3.6 Plus
AA Agentic Index	27.5	29
AA Intelligence	37.7	40.5
AA-LCR	73.3	72.3
AA-Omniscience	-20	2.6
aime_2026	94.1	95.3
Artificial Analysis Coding Index	53.7	54.5
C-Eval	91.4	93.3
Claw Eval (pass@3)	60.6	58.7
critpt	1.1	2.9
ERQA	62.5	65.7
gdpval	32	32
GPQA Diamond	87.8	90.4
HLE	24	28.8
HMMT 2025	90.7	94.6
hmmt_feb_2026	84.3	87.8
hmmt_nov_2025	90.7	94.6
IFBench	69.1	75.2
imo_answer_bench	80.8	83.8
livebench_agentic_coding	39.3	41.4
livebench_coding	71.8	78.2
livebench_data_analysis	70.4	69.9
livebench_instruction_following	53.2	58.3
livebench_language	63.3	75
livebench_math	79.9	83.7
livebench_reasoning	70.3	75.8
LiveCodeBench v6	83.9	87.1
mathvision	85.1	88
MLVU	86.6	86.7
mmlu_redux	93.5	94.5
MMLU-Pro	86.2	88.5
MMMU	82.9	86
MMMU-Pro	75.8	78.8
MMStar	81.4	83.3
nl2repo	36.2	37.9
OmniDocBench 1.5	89.4	91.2
OmniScience Accuracy	19.6	26.4
OmniScience Non-Hallucination	50.7	68
OSWorld-Verified	63.9	62.5
RealWorldQA	84.1	85.4
scicode	39.8	40.7
SimpleVQA	56.1	0.7
supergpqa	66	71.6
SWE-bench Multilingual	71.3	73.8
SWE-bench Pro	53.5	56.6
SWE-bench Verified	77.2	78.8
TauBench V3 - Banking	16.7	20.8
Terminal-Bench 2.0	59.3	61.6
Terminal-Bench 2.1	60.7	61.4
Terminal-Bench Hard	34.8	43.9
VideoMMMU	84.4	84
τ²-Bench Telecom (AA run)	94.2	97.7
τ³-Bench	15.3	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.