Which is better, Claude Opus 4.8 or Qwen3.7 Max?

Across 55 shared benchmarks tracked on Vector Wire, Claude Opus 4.8 leads on 40 and Qwen3.7 Max leads on 14 — 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 Qwen3.7 Max

AnthropicvsAlibaba55 shared benchmarks40–14 head-to-head

Benchmark	Claude Opus 4.8	Qwen3.7 Max
AA Agentic Index	49.4	30.9
AA Intelligence	57.3	46.7
AA-LCR	73	74.7
AA-Omniscience	28.8	13.5
Agents' Last Exam (Pass / Score)	45.1	31.1
aime_2026	100	97
AndroidBench	69.8	56.5
arena_elo	1482	1474
arena_text_factuality	1464	1479
Artificial Analysis Coding Index	74.3	66
coding_arena_elo	1539	1517
critpt	20.9	13.4
DeepSWE	59	18
DeepSWE 1.1	59	21.6
Finance Agent v2	53.9	48.4
gdpval	54.2	38.5
GPQA Diamond	93.6	92.4
HealthBench	59.3	54.5
HLE	57.9	41.4
HLE (with tools)	57.9	53.5
hmmt_feb_2026	96.7	97.1
hmmt_nov_2025	96.5	95
IFBench	62.2	80.5
imo_answer_bench	83.5	90
include	87.6	86.2
JobBench	48.4	31.3
livebench	77.2	74.3
livebench_agentic_coding	50.5	43.6
livebench_coding	81.8	74.2
livebench_data_analysis	66	71.8
livebench_instruction_following	72	74
livebench_language	79.7	79.7
livebench_math	94.3	85.3
livebench_reasoning	89.2	83.3
longbench_v2	69.1	65.3
MCP Atlas	83.6	76.4
MLS Bench Lite	42.8	31.7
MRCR v2 256K (8-needle)	83.2	86.7
nl2repo	69.7	47.2
OmniScience Accuracy	48.8	31.1
OmniScience Non-Hallucination	60.7	74.4
PaperBench	80.3	64.8
scicode	53.5	53.5
simplebench	64.8	70.4
simpleqa_verified	39.5	58.5
SWE-bench Multilingual	84.4	78.3
SWE-bench Pro	69.2	60.6
SWE-bench Verified	88.6	80.4
TauBench V3 - Banking	34.2	11.8
Terminal-Bench 2.0	74.6	69.7
Terminal-Bench 2.1	85	75
Terminal-Bench Hard	58.3	50.8
WideSearch	72.9	75.2
τ²-Bench Telecom (AA run)	94.4	94.7
τ³-Bench	27.6	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.