Which is better, GPT-5.5 or Qwen3.7 Max?

Across 47 shared benchmarks tracked on Vector Wire, GPT-5.5 leads on 39 and Qwen3.7 Max leads on 8 — 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 —

GPT-5.5 vs Qwen3.7 Max

OpenAIvsAlibaba47 shared benchmarks39–8 head-to-head

Benchmark	GPT-5.5	Qwen3.7 Max
AA Agentic Index	47.4	30.9
AA Intelligence	56.3	46.7
AA-LCR	79	74.7
AA-Omniscience	20.5	13.5
aime_2026	100	97
arena_elo	1482	1474
arena_text_factuality	1482	1479
Artificial Analysis Coding Index	74.9	66
coding_arena_elo	1457	1517
critpt	27.1	13.4
DeepSWE	70	18
DeepSWE 1.1	67	21.6
Finance Agent v2	51.8	48.4
gdpval	49.5	38.5
GPQA Diamond	93.6	92.4
HealthBench	56.5	54.5
HLE	52.2	41.4
HLE (with tools)	52.2	53.5
hmmt_feb_2026	98.5	97.1
hmmt_nov_2025	96.5	95
IFBench	75.9	80.5
itbenchSre	45.8	42.5
JobBench	38.3	31.3
livebench	80.7	74.3
livebench_agentic_coding	54	43.6
livebench_coding	82.2	74.2
livebench_data_analysis	81.6	71.8
livebench_instruction_following	70.7	74
livebench_language	87.4	79.7
livebench_math	95.9	85.3
livebench_reasoning	89.7	83.3
MCP Atlas	82.8	76.4
MLS Bench Lite	35.5	31.7
nl2repo	50.7	47.2
OmniScience Accuracy	58	31.1
OmniScience Non-Hallucination	12	74.4
scicode	56.1	53.5
simplebench	69	70.4
simpleqa_verified	63.1	58.5
SWE-bench Pro	58.6	60.6
SWE-bench Verified	80.6	80.4
TauBench V3 - Banking	39	11.8
Terminal-Bench 2.0	82.7	69.7
Terminal-Bench 2.1	84.3	75
Terminal-Bench Hard	60.6	50.8
τ²-Bench Telecom (AA run)	93.9	94.7
τ³-Bench	31.3	10.9

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.