Which is better, GLM-5.2 Full Open Source or Kimi K2.5?

Across 46 shared benchmarks tracked on Vector Wire, GLM-5.2 Full Open Source leads on 40 and Kimi K2.5 leads on 6 — GLM-5.2 Full Open Source leads on more. Strengths differ by axis; the table below shows every head-to-head number with provenance.

SEEDRefresh Models —Deals —Regulatory —Sources —

GLM-5.2 Full Open Source vs Kimi K2.5

Z.aivsMoonshot46 shared benchmarks40–6 head-to-head

Benchmark	GLM-5.2 Full Open Source	Kimi K2.5
AA Agentic Index	45.7	52.8
AA Intelligence	52.6	36
AA-LCR	76.7	73
AA-Omniscience	4.4	-7.3
aime_2026	99.2	95.8
apexAgents	33.7	11.5
arc_agi_1	77	65.3
ARC-AGI-2	22.8	11.8
arena_elo	1471	1450
arena_text_factuality	1460	1445
Artificial Analysis Coding Index	68.8	46.8
coding_arena_elo	1582	1436
critpt	20.9	3.1
FORTRESS (Adversarial)	71.3	54.1
FORTRESS (Benign)	90	98.3
gdpval	50.3	38.3
GDPval-AA v2	1514	1009
Global-MMLU-Lite	89.2	84
GPQA Diamond	91.2	87.9
HLE	54.7	50.2
HLE (with tools)	54.7	51.8
HMMT 2025	94.4	95.4
hmmt_feb_2026	92.5	87.1
hmmt_nov_2025	94.4	91.1
IFBench	73.3	70.2
imo_answer_bench	91	81.8
MCP Atlas	82.6	64
nl2repo	48.9	32
OmniScience Accuracy	24.3	35.2
OmniScience Non-Hallucination	73.7	50
ResearchRubrics	71.1	59.5
scicode	50.5	49
simplebench	58.8	46.8
simpleqa_verified	38.1	36.9
strongreject	98.5	99.5
SWE-bench Pro	62.1	53.8
SWEBench Pro (Public)	62.1	50.7
Tau 3 Banking	26.8	14.2
TauBench V3 - Banking	34.6	14.2
Terminal Bench 2.1 (Best Harness)	82.7	51.3
Terminal-Bench 2.1	82.7	45.7
Terminal-Bench Hard	50.8	34.8
tool_decathlon	48.2	27.8
toolathlon	48.2	27.8
τ²-Bench Telecom (AA run)	99.1	95.9
τ³-Bench	26.8	66

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.