Which is better, Gemini 2.5 Pro or GPT-5.2?

Across 52 shared benchmarks tracked on Vector Wire, Gemini 2.5 Pro leads on 13 and GPT-5.2 leads on 39 — GPT-5.2 leads on more. Strengths differ by axis; the table below shows every head-to-head number with provenance.

SEEDRefresh Models —Deals —Regulatory —Sources —

Gemini 2.5 Pro vs GPT-5.2

GooglevsOpenAI52 shared benchmarks13–39 head-to-head

Benchmark	Gemini 2.5 Pro	GPT-5.2
AA Agentic Index	7.2	60.2
AA Intelligence	27	43.3
AA-LCR	66	79.3
AA-Omniscience	-14.3	0.3
AIME 2025	88.3	100
arc_agi_1	41	86.2
ARC-AGI-2	4.9	52.9
arena_elo	1446	1437
arena_text_factuality	1462	1442
arena_vision	1246	1245
Artificial Analysis Coding Index	46.7	48.7
browsecomp	9.9	65.8
browsecomp_zh	32.2	76.1
coding_arena_elo	1224	1418
critpt	2.6	11.6
Fortress	54.9	17.5
frontiermath_tier_4	4.2	18.8
gdpval	8.5	48.3
GPQA Diamond	84.4	92.4
HLE	22.5	45.5
HLE (with tools)	28.4	45.5
HMMT 2025	65.7	99.4
hmmt_feb_2025	82.5	100
hmmt_nov_2025	80	99.2
IFBench	49	75.4
LiveCodeBench	82.7	89
matharena_visual_math_overall	77.2	86.5
mathvision	73.3	83
MathVista	83.9	82.8
MMLU-Pro	86	87
MMMU-Pro	74.9	79.5
OCRBench	85.9	80.7
ocrbench_v2	59.3	50.5
OmniScience Accuracy	39	44.3
OmniScience Non-Hallucination	12.6	38.4
PropensityBench	79	34.4
scicode	43	52.1
SEAL VISTA	50.8	46.6
simplebench	62.4	45.8
simpleqa_verified	56	38.9
swe_bench_bash	53.6	72.8
SWE-bench Verified	67.2	80
TauBench V3 - Banking	9.7	11.1
Terminal-Bench Hard	26.5	62.2
vectara_answer_rate	99.1	100
vectara_avg_summary_length	106.4	186.3
vectara_factual_consistency	93	89.2
vectara_hallucination_rate ↓	7	10.8
Video-MME	84.8	86
τ²-Bench	54	85.5
τ²-Bench Telecom (AA run)	54.1	98.7
τ³-Bench	9.3	12

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.