Which is better, Claude Opus 4.5 or Qwen3.5 122B A10B?

Refresh Models —Deals —Regulatory —Sources —

Questions this page answers

Which is better, Claude Opus 4.5 or Qwen3.5 122B A10B?: Across 51 shared benchmarks, Claude Opus 4.5 scores higher on 31 and Qwen3.5 122B A10B on 20. The widest gap is critpt, where Claude Opus 4.5 scores 4.6 against 0.9. Qwen3.5 122B A10B is the cheaper of the two on tracked API pricing ($0.40 against $5.00 per million input tokens).

Claude Opus 4.5 vs Qwen3.5 122B A10B

Across 51 shared benchmarks, Claude Opus 4.5 scores higher on 31 and Qwen3.5 122B A10B on 20. The widest gap is critpt, where Claude Opus 4.5 scores 4.6 against 0.9. Qwen3.5 122B A10B is the cheaper of the two on tracked API pricing ($0.40 against $5.00 per million input tokens).

AnthropicvsAlibaba51 shared benchmarks31–20 head-to-head

Benchmark	Claude Opus 4.5	Qwen3.5 122B A10B
AA Agentic Index	59.6	21.3
AA Intelligence	41.9	32.8
AA-LCR	76	70.3
AA-Omniscience	14	-41.5
Artificial Analysis Coding Index	47.8	45.7
browsecomp	37	63.8
browsecomp_zh	62.4	69.9
C-Eval	92.2	91.9
CC-OCR	76.9	81.8
coding_arena_elo	1495	1358
critpt	4.6	0.9
DynaMath	79.7	85.9
ERQA	46.8	62
gdpval	47.3	24.3
GPQA Diamond	87	86.6
HLE	30.8	47.5
IFBench	58	76.1
LiveCodeBench v6	84.8	78.9
longbench_v2	64.4	60.2
mathvision	77.1	86.2
MathVista	80.2	87.4
MLVU	81.7	87.3
mmlu_redux	95.6	94
MMLU-Pro	90	86.7
mmmlu	90.8	86.7
MMMU	80.7	83.9
MMMU-Pro	74	76.9
MMStar	73.2	82.9
MMVU	77.3	74.7
multichallenge	59	61.5
MV-Bench	67.2	76.6
OCRBench	86.5	92.1
OmniDocBench 1.5	87.7	89.8
OmniScience Accuracy	46.6	24.4
OmniScience Non-Hallucination	39	12.9
OSWorld-Verified	66.3	58
RealWorldQA	77	85.1
scicode	50	42
Seal-0	47.7	44.1
SimpleVQA	69.7	0.6
supergpqa	70.6	67.1
SWE-bench Verified	81.5	72
Terminal-Bench 2.0	59.3	49.4
Terminal-Bench Hard	47	31.1
vectara_answer_rate	98.7	99.8
vectara_avg_summary_length	114.5	86.4
vectara_factual_consistency	89.1	88.8
vectara_hallucination_rate ↓	10.9	11.2
VideoMMMU	84.4	82
ZeroBench	3	0.1
τ²-Bench Telecom (AA run)	89.5	93.6

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. Quoted rates are the price-setter row we currently track for each model — its direct or vendor-official listing where one exists (anthropic-official, direct), otherwise the lowest tracked offer.