Which is better, GLM-5.1 or Qwen3.5 397B A17B?

Across 44 shared benchmarks tracked on Vector Wire, GLM-5.1 leads on 33 and Qwen3.5 397B A17B leads on 11 — GLM-5.1 leads on more. Strengths differ by axis; the table below shows every head-to-head number with provenance.

SEEDRefresh Models —Deals —Regulatory —Sources —

GLM-5.1 vs Qwen3.5 397B A17B

Z.aivsAlibaba44 shared benchmarks33–11 head-to-head

Benchmark	GLM-5.1	Qwen3.5 397B A17B
AA Agentic Index	66	53.3
AA Intelligence	41	34.3
AA-LCR	68	72.7
AA-Omniscience	0.8	-30.8
AgentWorldBench - Android	59.1	54.9
AgentWorldBench - MCP	67.6	68.3
AgentWorldBench - OS	59.1	60.9
AgentWorldBench - Overall	51.3	54.7
AgentWorldBench - Search	22.5	30.8
AgentWorldBench - SWE	52.1	64.4
AgentWorldBench - Terminal	47.3	55.3
AgentWorldBench - Web	51.5	48.5
aime_2026	95.8	94.2
arena_elo	1468	1442
arena_text_factuality	1458	1445
Artificial Analysis Coding Index	55.8	48.2
browsecomp	79.3	69
coding_arena_elo	1510	1399
critpt	4.6	1.7
gdpval	49.5	35.8
GPQA Diamond	86.8	89.3
HLE	52.3	29
HLE (with tools)	52.3	48.3
HMMT 2025	94	92.7
hmmt_feb_2026	89.4	87.9
hmmt_nov_2025	94	92.7
IFBench	76.3	78.8
imo_answer_bench	83.8	80.9
itbenchSre	40.3	34.1
MMLU-Pro	86	87.8
nl2repo	42.7	32.2
OmniScience Accuracy	25.2	30.8
OmniScience Non-Hallucination	70.1	17.3
scicode	43.8	42
simpleqa_verified	38.1	26
SWE-bench Multilingual	73.3	69.3
SWE-bench Pro	58.4	50.9
TauBench V3 - Banking	13.6	13.4
Terminal-Bench 2.0	69	52.5
Terminal-Bench 2.1	63.5	51.3
Terminal-Bench Hard	43.2	40.9
toolathlon	40.7	38.3
τ²-Bench Telecom (AA run)	97.7	95.6
τ³-Bench	70.6	13.4

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.