Which is better, GPT-5.6 Sol or Qwen3.7 Max?

Across 45 shared benchmarks tracked on Vector Wire, GPT-5.6 Sol leads on 39 and Qwen3.7 Max leads on 6 — 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 —

GPT-5.6 Sol vs Qwen3.7 Max

OpenAIvsAlibaba45 shared benchmarks39–6 head-to-head

Benchmark	GPT-5.6 Sol	Qwen3.7 Max
AA Agentic Index	57.8	30.9
AA Intelligence	61	46.7
AA-LCR	77.7	74.7
AA-Omniscience	22	13.5
Agents' Last Exam (Pass / Score)	53.6	31.1
aime_2026	99.9	97
AndroidBench	74	56.5
arena_elo	1482	1474
arena_text_factuality	1474	1479
Artificial Analysis Coding Index	78.3	66
coding_arena_elo	1619	1517
critpt	32.3	13.4
DeepSWE	73	18
DeepSWE 1.1	73	21.6
Finance Agent v2	53.8	48.4
gdpval	61.1	38.5
GPQA Diamond	94.6	92.4
HealthBench	57	54.5
HLE	49.5	41.4
HLE (with tools)	58	53.5
IFBench	72.7	80.5
itbenchSre	56.2	42.5
JobBench	45.4	31.3
livebench_agentic_coding	56.2	43.6
livebench_coding	83.9	74.2
livebench_data_analysis	79.8	71.8
livebench_instruction_following	71.8	74
livebench_language	87.7	79.7
livebench_math	96.2	85.3
livebench_reasoning	91.7	83.3
longbench_v2	67.1	65.3
MCP Atlas	83.6	76.4
MLS Bench Lite	46.2	31.7
MRCR v2 256K (8-needle)	93.8	86.7
OmniScience Accuracy	59.4	31.1
OmniScience Non-Hallucination	10.6	74.4
PaperBench	90.5	64.8
scicode	56.9	53.5
simplebench	64.8	70.4
simpleqa_verified	71.6	58.5
SWE-bench Pro	64.6	60.6
TauBench V3 - Banking	44.3	11.8
Terminal-Bench 2.1	89.5	75
Terminal-Bench Hard	65.9	50.8
τ²-Bench Telecom (AA run)	85.1	94.7

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.