Which is better, Claude Opus 4.7 or Kimi K2.6?

Across 54 shared benchmarks tracked on Vector Wire, Claude Opus 4.7 leads on 42 and Kimi K2.6 leads on 12 — Claude Opus 4.7 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 4.7 vs Kimi K2.6

AnthropicvsMoonshot54 shared benchmarks42–12 head-to-head

Benchmark	Claude Opus 4.7	Kimi K2.6
AA Agentic Index	64.6	58.7
AA Intelligence	55	45.1
AA-LCR	75.3	76.7
AA-Omniscience	27.3	6.4
aime_2026	95.8	96.4
arena_elo	1494	1461
arena_text_factuality	1474	1455
arena_vision	1304	1264
Artificial Analysis Coding Index	73.6	61.8
browsecomp	79.8	86.3
charxiv_rq	82.1	86.7
coding_arena_elo	1557	1509
critpt	12	8
Finance Agent v2	51.5	44.9
frontiermath_tier_4	22.9	14.6
gdpval	58.6	41.4
GDPval-AA (Elo)	1753	1482
GPQA Diamond	94.2	91.1
HiL-Bench	41.7	18.7
HLE	54.7	37.5
HLE (with tools)	54.7	54
hmmt_feb_2026	93.9	94.7
IFBench	58.6	76
itbenchSre	46.7	31.2
livebench	76.9	72.2
livebench_agentic_coding	50.7	46.9
livebench_coding	82.1	78.6
livebench_data_analysis	78.3	65.1
livebench_instruction_following	66.7	64.4
livebench_language	77.9	75.1
livebench_math	92.8	84.3
livebench_reasoning	87.2	79.4
MCP Atlas	79.1	68.1
MMMU-Pro	78.8	80.1
OmniScience Accuracy	48.9	32.8
OmniScience Non-Hallucination	57.7	60.7
OSWorld-Verified	82.8	73.1
scicode	54.5	53.5
simpleqa_verified	50.6	38.7
SWE-bench Multilingual	80.5	76.7
SWE-bench Pro	64.3	58.6
SWE-bench Verified	87.6	80.2
TauBench V3 - Banking	34.6	23.3
Terminal-Bench 2.0	69.4	66.7
Terminal-Bench 2.1	83.1	65.9
Terminal-Bench Hard	54.5	43.9
toolathlon	59.3	50
usamo_2026	69.3	51.2
vectara_answer_rate	98	99.7
vectara_avg_summary_length	149.1	116.7
vectara_factual_consistency	88	89.2
vectara_hallucination_rate ↓	12	10.8
τ²-Bench Telecom (AA run)	88.6	95.9
τ³-Bench	28.9	20.6

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.