Which is better, GLM-5.2 Full Open Source or GPT-5.5?

Across 58 shared benchmarks tracked on Vector Wire, GLM-5.2 Full Open Source leads on 14 and GPT-5.5 leads on 44 — GPT-5.5 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 GPT-5.5

Z.aivsOpenAI58 shared benchmarks14–44 head-to-head

Benchmark	GLM-5.2 Full Open Source	GPT-5.5
AA Agentic Index	45.7	47.4
AA Intelligence	52.6	56.3
AA-LCR	76.7	79
AA-Omniscience	4.4	20.5
Agents' Last Exam	23.8	26.6
aime_2026	99.2	100
apex_agents	35.6	38.5
apexAgents	33.7	37.7
arc_agi_1	77	95
ARC-AGI-2	22.8	85
arena_elo	1471	1482
arena_text_factuality	1460	1482
Artificial Analysis Coding Index	68.8	74.9
coding_arena_elo	1582	1457
critpt	20.9	27.1
DeepSWE	46.2	70
DeepSWE 1.1	44	67
gdpval	50.3	49.5
GDPval-AA v2	1514	1509
GDPval-AA v2 (Elo)	1510	1491
GPQA Diamond	91.2	93.6
HiL-Bench	43.7	39.7
HLE	54.7	52.2
HLE (with tools)	54.7	52.2
hmmt_feb_2026	92.5	98.5
hmmt_nov_2025	94.4	96.5
IFBench	73.3	75.9
itbenchSre	42.7	45.8
JobBench	43.4	38.3
livebench_agentic_coding	51.8	54
livebench_coding	79.7	82.2
livebench_data_analysis	73.7	81.6
livebench_instruction_following	62.3	70.7
livebench_language	76.2	87.4
livebench_math	89.8	95.9
livebench_reasoning	78.6	89.7
MCP Atlas	82.6	82.8
MLS Bench Lite	40.4	35.5
nl2repo	48.9	50.7
officeqa_pro	41.4	60.9
OmniScience Accuracy	24.3	58
OmniScience Non-Hallucination	73.7	12
PostTrainBench	34.3	28.4
Program Bench	63.7	70.8
ResearchRubrics	71.1	64
scicode	50.5	56.1
simplebench	58.8	69
simpleqa_verified	38.1	63.1
SpreadsheetBench 2	28.1	29.1
SWE-bench Pro	62.1	58.6
SWE-Marathon	13	14
TauBench V3 - Banking	34.6	39
Terminal-Bench 2.1	82.7	84.3
Terminal-Bench Hard	50.8	60.6
tool_decathlon	48.2	55.6
toolathlon	48.2	55.6
τ²-Bench Telecom (AA run)	99.1	93.9
τ³-Bench	26.8	31.3

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.