Which is better, Kimi K2.5 or Qwen3.6 Plus?

Across 50 shared benchmarks tracked on Vector Wire, Kimi K2.5 leads on 17 and Qwen3.6 Plus leads on 33 — 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 —

Kimi K2.5 vs Qwen3.6 Plus

MoonshotvsAlibaba50 shared benchmarks17–33 head-to-head

Benchmark	Kimi K2.5	Qwen3.6 Plus
AA Agentic Index	52.8	29
AA Intelligence	36	40.5
AA-LCR	73	72.3
AA-Omniscience	-7.3	2.6
aime_2026	95.8	95.3
arena_elo	1450	1444
arena_text_factuality	1445	1438
Artificial Analysis Coding Index	46.8	54.5
coding_arena_elo	1436	1460
critpt	3.1	2.9
frontiermath_tier_4	4.2	8.3
gdpval	38.3	32
GPQA Diamond	87.9	90.4
HLE	50.2	28.8
HLE (with tools)	51.8	50.6
HMMT 2025	95.4	94.6
hmmt_feb_2026	87.1	87.8
hmmt_nov_2025	91.1	94.6
IFBench	70.2	75.2
imo_answer_bench	81.8	83.8
livebench	69.1	70.8
LiveCodeBench v6	85	87.1
longbench_v2	61	62
mathvision	84.2	88
MCP Atlas	64	74.1
mcpmark	29.5	48.2
MMLU-Pro	87.1	88.5
MMMU-Pro	78.5	78.8
nl2repo	32	37.9
OmniDocBench 1.5	88.8	91.2
OmniScience Accuracy	35.2	26.4
OmniScience Non-Hallucination	50	68
OSWorld-Verified	63.3	62.5
scicode	49	40.7
simpleqa_verified	36.9	49.1
SimpleVQA	71.2	0.7
SWE-bench Multilingual	73	73.8
SWE-bench Pro	53.8	56.6
SWE-bench Verified	76.8	78.8
TauBench V3 - Banking	14.2	20.8
Terminal-Bench 2.0	50.8	61.6
Terminal-Bench 2.1	45.7	61.4
Terminal-Bench Hard	34.8	43.9
tool_decathlon	27.8	39.8
toolathlon	27.8	39.8
Video-MME	87.4	84.2
VideoMMMU	86.6	84
WideSearch	79	74.3
τ²-Bench Telecom (AA run)	95.9	97.7
τ³-Bench	66	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.