Which is better, Qwen3.6 35B A3B or Qwen3.6 Plus?

Across 49 shared benchmarks tracked on Vector Wire, Qwen3.6 35B A3B leads on 1 and Qwen3.6 Plus leads on 48 — 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 35B A3B vs Qwen3.6 Plus

AlibabavsAlibaba49 shared benchmarks1–48 head-to-head

Benchmark	Qwen3.6 35B A3B	Qwen3.6 Plus
AA Agentic Index	21.6	29
AA Intelligence	32.1	40.5
AA-LCR	66.7	72.3
AA-Omniscience	-22.2	2.6
AI2D	92.7	94.4
aime_2026	92.7	95.3
Artificial Analysis Coding Index	41.9	54.5
C-Eval	90	93.3
Claw Eval (pass@3)	50	58.7
critpt	0.3	2.9
DeepPlanning	25.9	41.5
gdpval	27.8	32
GPQA Diamond	86	90.4
HLE	22.2	28.8
HMMT 2025	89.1	94.6
hmmt_feb_2026	83.6	87.8
hmmt_nov_2025	89.1	94.6
IFBench	64.4	75.2
imo_answer_bench	78.9	83.8
LiveCodeBench v6	80.4	87.1
MCP Atlas	62.8	74.1
mcpmark	37	48.2
MLVU	86.2	86.7
mmlu_redux	93.3	94.5
MMLU-Pro	85.6	88.5
MMMU	81.7	86
MMMU-Pro	75.3	78.8
nl2repo	29.4	37.9
OmniDocBench 1.5	89.9	91.2
OmniScience Accuracy	18.8	26.4
OmniScience Non-Hallucination	49.5	68
RealWorldQA	85.3	85.4
scicode	35.8	40.7
SimpleVQA	58.9	0.7
supergpqa	64.7	71.6
SWE-bench Multilingual	67.2	73.8
SWE-bench Pro	49.5	56.6
SWE-bench Verified	73.4	78.8
TauBench V3 - Banking	9.3	20.8
Terminal-Bench 2.0	51.5	61.6
Terminal-Bench 2.1	44.9	61.4
Terminal-Bench Hard	34.8	43.9
tool_decathlon	26.9	39.8
toolathlon	26.9	39.8
VideoMMMU	83.7	84
VITA-Bench	35.6	44.3
WideSearch	60.1	74.3
τ²-Bench Telecom (AA run)	95.3	97.7
τ³-Bench	67.2	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.