BenchLM 语言模型排行榜

按 BenchAlign 综合分比较语言模型,查看推理、编程与多模态表现。

评测来源 BenchLM采集 / 更新 查看原榜评分方法

BenchAlign

212 个模型
BenchLM · BenchAlign
排名模型 / 开发机构BenchAlign推理编程多模态详情
01Claude Fable 5.2Anthropic92.8BenchAlign推理91.0编程89.7多模态89.9推理91.0编程89.7多模态89.9详情
02GPT-6.2 AstraOpenAI89.3BenchAlign推理—编程—多模态—推理—编程—多模态—来源
03Claude Opus 5.5Anthropic85.8BenchAlign推理80.5编程82.9多模态84.2推理80.5编程82.9多模态84.2详情
04Gemini 4 CarbonGoogle85.0BenchAlign推理84.0编程83.4多模态—推理84.0编程83.4多模态—详情
05GPT-6.2 SolOpenAI86.1BenchAlign推理—编程—多模态—推理—编程—多模态—来源
06GPT-6 AstraOpenAI83.0BenchAlign推理80.7编程78.4多模态78.5推理80.7编程78.4多模态78.5详情
07GPT-6.1 SolOpenAI82.0BenchAlign推理80.0编程72.9多模态77.6推理80.0编程72.9多模态77.6详情
08Claude Sonnet 5.5Anthropic80.6BenchAlign推理76.0编程80.5多模态77.0推理76.0编程80.5多模态77.0详情
09Gemini 4 ArgonGoogle79.3BenchAlign推理77.2编程72.1多模态—推理77.2编程72.1多模态—详情
10Claude Fable 5.1Anthropic78.6BenchAlign推理75.9编程73.7多模态78.9推理75.9编程73.7多模态78.9详情
11Claude Fable 5Anthropic76.2BenchAlign推理71.5编程67.6多模态73.7推理71.5编程67.6多模态73.7详情
12Claude Opus 5Anthropic76.0BenchAlign推理73.7编程66.8多模态74.5推理73.7编程66.8多模态74.5详情
13Gemini 4 HydrogenGoogle77.9BenchAlign推理—编程75.8多模态—推理—编程75.8多模态—来源
14GPT-6 SolOpenAI75.3BenchAlign推理71.7编程66.7多模态73.5推理71.7编程66.7多模态73.5详情
15GPT-6.1 TerraOpenAI77.1BenchAlign推理74.9编程69.4多模态76.0推理74.9编程69.4多模态76.0详情
16GPT-5.6 SolOpenAI70.9BenchAlign推理72.6编程62.2多模态73.3推理72.6编程62.2多模态73.3详情
17Kimi K3.1 ProMoonshot AI71.3BenchAlign推理69.2编程61.3多模态71.8推理69.2编程61.3多模态71.8详情
18GPT-5.5 ProOpenAI70.8BenchAlign推理74.8编程—多模态—推理74.8编程—多模态—详情
19Muse Wonder FullMeta70.6BenchAlign推理75.9编程70.2多模态—推理75.9编程70.2多模态—详情
20Kimi K3Moonshot AI66.7BenchAlign推理65.3编程56.1多模态66.8推理65.3编程56.1多模态66.8详情
21GPT-5.6 TerraOpenAI66.3BenchAlign推理68.2编程57.3多模态67.1推理68.2编程57.3多模态67.1详情
22Claude Opus 4.8Anthropic65.9BenchAlign推理63.5编程55.5多模态64.8推理63.5编程55.5多模态64.8详情
23GPT-5.5OpenAI65.8BenchAlign推理65.8编程57.9多模态66.5推理65.8编程57.9多模态66.5详情
24Claude Opus 4.7Anthropic65.6BenchAlign推理58.9编程56.3多模态62.1推理58.9编程56.3多模态62.1详情
25GPT-5.2 ProOpenAI65.2BenchAlign推理—编程—多模态—推理—编程—多模态—详情
26Grok 4.6xAI64.7BenchAlign推理64.3编程54.6多模态—推理64.3编程54.6多模态—详情
27Ling 3.1 FlashInclusionAI64.6BenchAlign推理62.1编程57.3多模态—推理62.1编程57.3多模态—详情
28GPT-5.4 ProOpenAI66.5BenchAlign推理72.3编程—多模态—推理72.3编程—多模态—详情
29Gemini 3.7 FlashGoogle64.4BenchAlign推理62.1编程54.5多模态68.3推理62.1编程54.5多模态68.3详情
30Gemini 3.8 FlashGoogle64.4BenchAlign推理64.2编程55.1多模态68.8推理64.2编程55.1多模态68.8详情
31GPT-5.4OpenAI64.1BenchAlign推理61.8编程48.8多模态63.3推理61.8编程48.8多模态63.3详情
32Muse WonderMeta66.1BenchAlign推理69.7编程64.8多模态—推理69.7编程64.8多模态—详情
33DeepSeek V4.1 Flash 1017DeepSeek63.9BenchAlign推理60.4编程59.8多模态64.0推理60.4编程59.8多模态64.0详情
34Muse Spark 1.2Meta63.7BenchAlign推理59.8编程49.6多模态—推理59.8编程49.6多模态—详情
35Qwen3.8 MaxAlibaba63.4BenchAlign推理60.1编程50.7多模态64.6推理60.1编程50.7多模态64.6详情
36Muse Spark 1.1Meta63.3BenchAlign推理58.8编程51.2多模态—推理58.8编程51.2多模态—详情
37Gemini 3.1 ProGoogle62.9BenchAlign推理56.5编程37.7多模态64.3推理56.5编程37.7多模态64.3详情
38Claude Haiku 5.5Anthropic62.9BenchAlign推理57.6编程58.0多模态59.5推理57.6编程58.0多模态59.5详情
39DeepSeek V4 ProDeepSeek62.7BenchAlign推理58.1编程51.4多模态—推理58.1编程51.4多模态—来源
40dots3-note PreviewDots Studio62.7BenchAlign推理—编程—多模态64.5推理—编程—多模态64.5详情
41Qwen4 MaxAlibaba64.4BenchAlign推理—编程—多模态—推理—编程—多模态—来源
42Claude Sonnet 5Anthropic62.3BenchAlign推理60.3编程51.6多模态59.7推理60.3编程51.6多模态59.7详情
43GPT-5.3 CodexOpenAI62.2BenchAlign推理58.0编程52.2多模态—推理58.0编程52.2多模态—详情
44Gemini 3.5 FlashGoogle61.8BenchAlign推理51.9编程49.8多模态67.6推理51.9编程49.8多模态67.6详情
45GPT-5.6 LunaOpenAI61.6BenchAlign推理57.5编程57.6多模态63.7推理57.5编程57.6多模态63.7详情
46Grok 4.5xAI61.5BenchAlign推理57.1编程49.1多模态—推理57.1编程49.1多模态—详情
47Ornith-1.5-397BOrnith AI61.3BenchAlign推理—编程53.9多模态—推理—编程53.9多模态—详情
48Claude Opus 4.6Anthropic61.3BenchAlign推理56.7编程45.5多模态55.3推理56.7编程45.5多模态55.3详情
49Gemini 3.6 FlashGoogle61.2BenchAlign推理55.5编程50.6多模态65.4推理55.5编程50.6多模态65.4详情
50GPT-6 LunaOpenAI61.1BenchAlign推理54.1编程51.8多模态62.5推理54.1编程51.8多模态62.5详情
显示更多已显示 50 / 212

保留来源评分口径,请在同一来源、同一任务内比较。

未公布的数据记为 —