Which is better, Claude Opus 4.8 or Qwen3.5 397B A17B?

Across 51 shared benchmarks tracked on Vector Wire, Claude Opus 4.8 leads on 43 and Qwen3.5 397B A17B leads on 8 — Claude Opus 4.8 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 4.8 vs Qwen3.5 397B A17B

AnthropicvsAlibaba51 shared benchmarks43–8 head-to-head

Benchmark	Claude Opus 4.8	Qwen3.5 397B A17B
AA Agentic Index	49.4	53.3
AA Intelligence	57.3	34.3
AA-LCR	73	72.7
AA-Omniscience	28.8	-30.8
AgentWorldBench - Android	61.5	54.9
AgentWorldBench - MCP	54.9	68.3
AgentWorldBench - OS	66.6	60.9
AgentWorldBench - Overall	56.6	54.7
AgentWorldBench - Search	35.1	30.8
AgentWorldBench - SWE	64.1	64.4
AgentWorldBench - Terminal	59.2	55.3
AgentWorldBench - Web	54.7	48.5
aime_2026	100	94.2
arena_elo	1482	1442
arena_text_factuality	1464	1445
arena_vision	1280	1249
Artificial Analysis Coding Index	74.3	48.2
browsecomp	88.5	69
charxiv_rq	80.5	80.8
coding_arena_elo	1539	1399
critpt	20.9	1.7
gdpval	54.2	35.8
GDPval-AA v2	1593	962
GPQA Diamond	93.6	89.3
HLE	57.9	29
HLE (with tools)	57.9	48.3
hmmt_feb_2026	96.7	87.9
hmmt_nov_2025	96.5	92.7
IFBench	62.2	78.8
imo_answer_bench	83.5	80.9
include	87.6	85.6
longbench_v2	69.1	63.2
MMMU-Pro	78.9	79
nl2repo	69.7	32.2
OmniScience Accuracy	48.8	30.8
OmniScience Non-Hallucination	60.7	17.3
scicode	53.5	42
simpleqa_verified	39.5	26
SWE-bench Multilingual	84.4	69.3
SWE-bench Pro	69.2	50.9
SWE-bench Verified	88.6	76.4
Tau 3 Banking	27.6	13.4
TauBench V3 - Banking	34.2	13.4
Terminal-Bench 2.0	74.6	52.5
Terminal-Bench 2.1	85	51.3
Terminal-Bench Hard	58.3	40.9
toolathlon	59.9	38.3
usamo_2026	96.7	36.3
WideSearch	72.9	74
τ²-Bench Telecom (AA run)	94.4	95.6
τ³-Bench	27.6	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.