Which is better, Claude Opus 4.6 or Qwen3.5 397B A17B?

Across 53 shared benchmarks tracked on Vector Wire, Claude Opus 4.6 leads on 43 and Qwen3.5 397B A17B leads on 10 — 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 Qwen3.5 397B A17B

AnthropicvsAlibaba53 shared benchmarks43–10 head-to-head

Benchmark	Claude Opus 4.6	Qwen3.5 397B A17B
AA Agentic Index	67.6	53.3
AA Intelligence	44.9	34.3
AA-LCR	74.3	72.7
AA-Omniscience	13.7	-30.8
AgentWorldBench - Android	61.7	54.9
AgentWorldBench - MCP	69.9	68.3
AgentWorldBench - OS	70.2	60.9
AgentWorldBench - Overall	57.8	54.7
AgentWorldBench - Search	29.3	30.8
AgentWorldBench - SWE	64.5	64.4
AgentWorldBench - Terminal	57.5	55.3
AgentWorldBench - Web	51.4	48.5
aime_2026	96.7	94.2
apexAgents	33	15.3
arena_elo	1497	1442
arena_text_factuality	1487	1445
arena_vision	1300	1249
Artificial Analysis Coding Index	48.1	48.2
browsecomp	86.8	69
charxiv_rq	69.1	80.8
coding_arena_elo	1545	1399
critpt	12.6	1.7
ERQA	40.8	67.5
gdpval	55.9	35.8
GPQA Diamond	91.3	89.3
GSM8K	0.6	66.7
HLE	53.1	29
HLE (with tools)	62.7	48.3
hmmt_feb_2026	96.2	87.9
hmmt_nov_2025	96.3	92.7
IFBench	62.5	78.8
imo_answer_bench	75.3	80.9
LiveCodeBench v6	88.8	83.6
mcpmark	56.7	46.1
MMLU-Pro	89.1	87.8
mmmlu	91.1	88.5
MMMU-Pro	77.3	79
multichallenge	56	67.6
nl2repo	49.8	32.2
OmniScience Accuracy	47	30.8
OmniScience Non-Hallucination	37.2	17.3
RealWorldQA	73.9	83.9
scicode	52	42
simpleqa_verified	46.5	26
SWE-bench Multilingual	77.8	69.3
SWE-bench Pro	57.3	50.9
SWE-bench Verified	80.8	76.4
Terminal-Bench 2.0	65.4	52.5
Terminal-Bench Hard	65.4	40.9
toolathlon	56.8	38.3
usamo_2026	66.2	36.3
τ²-Bench Telecom (AA run)	99.3	95.6
τ³-Bench	72.4	13.4

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.