Which is better, GLM-5.2 Full Open Source or GPT-5.6 Sol?

Across 59 shared benchmarks tracked on Vector Wire, GLM-5.2 Full Open Source leads on 5 and GPT-5.6 Sol leads on 53 — GPT-5.6 Sol 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-5.6 Sol

Z.aivsOpenAI59 shared benchmarks5–53 head-to-head

Benchmark	GLM-5.2 Full Open Source	GPT-5.6 Sol
AA Agentic Index	45.7	57.8
AA Intelligence	52.6	61
AA-LCR	76.7	77.7
AA-Omniscience	4.4	22
Agents' Last Exam	23.8	52.7
aime_2026	99.2	99.9
apex_agents	35.6	56.7
arc_agi_1	77	97.5
ARC-AGI-2	22.8	92.5
arena_elo	1471	1482
arena_text_factuality	1460	1474
Artificial Analysis Coding Index	68.8	78.3
coding_arena_elo	1582	1619
critpt	20.9	32.3
DeepSWE	46.2	73
DeepSWE 1.1	44	73
FORTRESS (Adversarial)	71.3	82.4
FORTRESS (Benign)	90	98.1
gdpval	50.3	61.1
GDPval-AA v2	1514	1748
GDPval-AA v2 (Elo)	1510	1736
Global-MMLU-Lite	89.2	91.8
GPQA Diamond	91.2	94.6
HiL-Bench	43.7	32.3
HLE	54.7	49.5
HLE (with tools)	54.7	58
IFBench	73.3	72.7
itbenchSre	42.7	56.2
JobBench	43.4	45.4
livebench_agentic_coding	51.8	56.2
livebench_coding	79.7	83.9
livebench_data_analysis	73.7	79.8
livebench_instruction_following	62.3	71.8
livebench_language	76.2	87.7
livebench_math	89.8	96.2
livebench_reasoning	78.6	91.7
MCP Atlas	82.6	83.6
MLS Bench Lite	40.4	46.2
officeqa_pro	41.4	63.2
OmniScience Accuracy	24.3	59.4
OmniScience Non-Hallucination	73.7	10.6
PostTrainBench	34.3	34.6
Program Bench	63.7	77.6
ResearchRubrics	71.1	73.8
scicode	50.5	56.9
simplebench	58.8	64.8
simpleqa_verified	38.1	71.6
SpreadsheetBench 2	28.1	32.4
strongreject	98.5	98.5
SWE-bench Pro	62.1	64.6
SWE-Marathon	13	39
SWEBench Pro (Public)	62.1	64.6
Tau 3 Banking	26.8	33
TauBench V3 - Banking	34.6	44.3
Terminal Bench 2.1 (Best Harness)	82.7	89.5
Terminal-Bench 2.1	82.7	89.5
Terminal-Bench Hard	50.8	65.9
toolathlon	48.2	58
τ²-Bench Telecom (AA run)	99.1	85.1

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.