Which is better, Gemma 4 31B or Qwen3.6 Plus?

Across 54 shared benchmarks tracked on Vector Wire, Gemma 4 31B leads on 5 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 —

Gemma 4 31B vs Qwen3.6 Plus

GooglevsAlibaba54 shared benchmarks5–49 head-to-head

Benchmark	Gemma 4 31B	Qwen3.6 Plus
AA Agentic Index	14.4	29
AA Intelligence	29.7	40.5
AA-LCR	68.3	72.3
AA-Omniscience	-47.9	2.6
AI2D	89	94.4
aime_2026	89.2	95.3
arena_elo	1451	1444
arena_text_factuality	1449	1438
Artificial Analysis Coding Index	43.4	54.5
C-Eval	82.6	93.3
Claw Eval (pass@3)	25	58.7
coding_arena_elo	1364	1460
critpt	1.4	2.9
DeepPlanning	24	41.5
ERQA	57.5	65.7
gdpval	15.5	32
GPQA Diamond	85.7	90.4
HLE	26.5	28.8
HLE (with tools)	26.5	50.6
hmmt_feb_2026	77.2	87.8
hmmt_nov_2025	87.5	94.6
IFBench	75.6	75.2
imo_answer_bench	74.5	83.8
LiveCodeBench v6	80	87.1
mathvision	85.6	88
MCP Atlas	57.2	74.1
mcpmark	18.1	48.2
mmlu_redux	93.7	94.5
MMLU-Pro	85.2	88.5
mmmlu	88.4	89.5
MMMU	80.4	86
MMMU-Pro	76.9	78.8
MMStar	77.3	83.3
nl2repo	15.5	37.9
OmniDocBench 1.5	80.1	91.2
OmniScience Accuracy	20	26.4
OmniScience Non-Hallucination	18.1	68
RealWorldQA	72.3	85.4
scicode	43.4	40.7
simpleqa_verified	9.6	49.1
SimpleVQA	52.9	0.7
supergpqa	65.7	71.6
SWE-bench Multilingual	51.7	73.8
SWE-bench Pro	35.7	56.6
SWE-bench Verified	52	78.8
TauBench V3 - Banking	14.8	20.8
Terminal-Bench 2.0	42.9	61.6
Terminal-Bench 2.1	43.4	61.4
Terminal-Bench Hard	36.4	43.9
tool_decathlon	21.2	39.8
VITA-Bench	43	44.3
WideSearch	35.2	74.3
τ²-Bench Telecom (AA run)	65.5	97.7
τ³-Bench	67.5	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.