Which is better, Kimi K2.6 or Qwen3.7 Max?

Across 46 shared benchmarks tracked on Vector Wire, Kimi K2.6 leads on 12 and Qwen3.7 Max leads on 34 — Qwen3.7 Max leads on more. Strengths differ by axis; the table below shows every head-to-head number with provenance.

SEEDRefresh Models —Deals —Regulatory —Sources —

Kimi K2.6 vs Qwen3.7 Max

MoonshotvsAlibaba46 shared benchmarks12–34 head-to-head

Benchmark	Kimi K2.6	Qwen3.7 Max
AA Agentic Index	58.7	30.9
AA Intelligence	45.1	46.7
AA-LCR	76.7	74.7
AA-Omniscience	6.4	13.5
aime_2026	96.4	97
arena_elo	1461	1474
arena_text_factuality	1455	1479
Artificial Analysis Coding Index	61.8	66
Claw Eval (pass@3)	80.9	65.2
coding_arena_elo	1509	1517
critpt	8	13.4
Finance Agent v2	44.9	48.4
gdpval	41.4	38.5
GPQA Diamond	91.1	92.4
HLE	37.5	41.4
HLE (with tools)	54	53.5
hmmt_feb_2026	94.7	97.1
IFBench	76	80.5
imo_answer_bench	86	90
itbenchSre	31.2	42.5
livebench	72.2	74.3
livebench_agentic_coding	46.9	43.6
livebench_coding	78.6	74.2
livebench_data_analysis	65.1	71.8
livebench_instruction_following	64.4	74
livebench_language	75.1	79.7
livebench_math	84.3	85.3
livebench_reasoning	79.4	83.3
LiveCodeBench v6	89.6	91.6
MCP Atlas	68.1	76.4
mcpmark	55.9	60.8
MMLU-Pro	87.1	89.6
OmniScience Accuracy	32.8	31.1
OmniScience Non-Hallucination	60.7	74.4
scicode	53.5	53.5
simpleqa_verified	38.7	58.5
SWE-bench Multilingual	76.7	78.3
SWE-bench Pro	58.6	60.6
SWE-bench Verified	80.2	80.4
TauBench V3 - Banking	23.3	11.8
Terminal-Bench 2.0	66.7	69.7
Terminal-Bench 2.1	65.9	75
Terminal-Bench Hard	43.9	50.8
WideSearch	80.8	75.2
τ²-Bench Telecom (AA run)	95.9	94.7
τ³-Bench	20.6	10.9

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.