Which is better, Claude Opus 4.6 or Qwen3.7 Max?

Across 44 shared benchmarks tracked on Vector Wire, Claude Opus 4.6 leads on 23 and Qwen3.7 Max leads on 21 — Claude Opus 4.6 leads on more. Strengths differ by axis; the table below shows every head-to-head number with provenance.

SEEDRefresh Models —Deals —Regulatory —Sources —

Claude Opus 4.6 vs Qwen3.7 Max

AnthropicvsAlibaba44 shared benchmarks23–21 head-to-head

Benchmark	Claude Opus 4.6	Qwen3.7 Max
AA Agentic Index	67.6	30.9
AA Intelligence	44.9	46.7
AA-LCR	74.3	74.7
AA-Omniscience	13.7	13.5
aime_2026	96.7	97
arena_elo	1497	1474
arena_text_factuality	1487	1479
Artificial Analysis Coding Index	48.1	66
coding_arena_elo	1545	1517
critpt	12.6	13.4
gdpval	55.9	38.5
GPQA Diamond	91.3	92.4
HLE	53.1	41.4
HLE (with tools)	62.7	53.5
hmmt_feb_2026	96.2	97.1
hmmt_nov_2025	96.3	95
IFBench	62.5	80.5
imo_answer_bench	75.3	90
livebench	76.3	74.3
livebench_agentic_coding	49	43.6
livebench_coding	78.2	74.2
livebench_data_analysis	69.9	71.8
livebench_instruction_following	63.3	74
livebench_language	83.3	79.7
livebench_math	89.3	85.3
livebench_reasoning	88.7	83.3
LiveCodeBench v6	88.8	91.6
MCP Atlas	76.8	76.4
mcpmark	56.7	60.8
MMLU-Pro	89.1	89.6
mmmlu	91.1	90.3
nl2repo	49.8	47.2
OmniScience Accuracy	47	31.1
OmniScience Non-Hallucination	37.2	74.4
scicode	52	53.5
simplebench	67.6	70.4
simpleqa_verified	46.5	58.5
SWE-bench Multilingual	77.8	78.3
SWE-bench Pro	57.3	60.6
SWE-bench Verified	80.8	80.4
Terminal-Bench 2.0	65.4	69.7
Terminal-Bench Hard	65.4	50.8
τ²-Bench Telecom (AA run)	99.3	94.7
τ³-Bench	72.4	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.