Skip to content
LMC Feed-Models, Papers, Benchmarks. Zero Fluff.Live

每周AI性能报告

本周 August 9, 2026

81 个模型上升， 216 个下降， 3 个不变，本周共追踪 300 个模型。

1

执行摘要

最大赢家

Claude Fable 5 (batch)

+377 个名次

当前排名 #2

最大输家

Seed 1.6 Flash

-65 个名次

当前排名 #297

新上榜

10

个新模型本周进入排名

2

每周变动

过去7天内排名绝对变动最大的10个模型。

模型评分7天变化排名
Claude Fable 5 (batch)Anthropic97.1+377#2
Claude Opus 4.7 (batch)Anthropic95.1+370#9
Claude Opus 4.8 (batch)Anthropic94.6+369#10
GPT-5.5 Pro (batch)OpenAI92.7+367#12
GPT-5.5 (batch)OpenAI92.7+365#14
Gemini 3.1 Pro Preview (batch)Google92.2+362#17
GPT-5.4 Pro (batch)OpenAI91.9+360#19
GPT-5.4 (batch)OpenAI91.9+358#21
GPT-5.2 Pro (batch)OpenAI90.5+352#27
GPT-5.2 (batch)OpenAI90.5+350#29
3

涨幅最大

本周排名提升最大的模型。

1
Claude Fable 5 (batch)Anthropic
+377
2
Claude Opus 4.7 (batch)Anthropic
+370
3
Claude Opus 4.8 (batch)Anthropic
+369
4
GPT-5.5 Pro (batch)OpenAI
+367
5
GPT-5.5 (batch)OpenAI
+365
6
Gemini 3.1 Pro Preview (batch)Google
+362
7
GPT-5.4 Pro (batch)OpenAI
+360
8
GPT-5.4 (batch)OpenAI
+358
9
GPT-5.2 Pro (batch)OpenAI
+352
10
GPT-5.2 (batch)OpenAI
+350
4

跌幅最大

本周排名下降最大的模型。

1
GPT-3.5 Turbo 16kOpenAI
-70
2
GPT-3.5 TurboOpenAI
-70
3
Nova Lite 1.0Amazon
-69
4
Nova Micro 1.0Amazon
-69
5
Nova Pro 1.0Amazon
-69
6
Qwen2.5 72B InstructAlibaba
-69
7
Mistral NemoMistral AI
-69
8
GPT-3.5 Turbo (older v0613)OpenAI
-69
9
Gemma 3n 4BGoogle
-67
10
Virtuoso Largearcee-ai
-67
5

新模型

10 个新模型本周进入排名。

模型评分排名
Claude Fable 5 (batch)Anthropic97.1#2
Claude Opus 4.7 (batch)Anthropic95.1#9
Claude Opus 4.8 (batch)Anthropic94.6#10
GPT-5.5 Pro (batch)OpenAI92.7#12
GPT-5.5 (batch)OpenAI92.7#14
Gemini 3.1 Pro Preview (batch)Google92.2#17
GPT-5.4 Pro (batch)OpenAI91.9#19
GPT-5.4 (batch)OpenAI91.9#21
GPT-5.2 Pro (batch)OpenAI90.5#27
GPT-5.2 (batch)OpenAI90.5#29
6

关注列表

处于脆弱状态可能进一步恶化的模型。需密切关注。

模型评分7天变化
Claude Opus 5 (Fast)Anthropic
95.1+171
Claude Opus 5Anthropic
95.1+171
GPT-5.5OpenAI
92.7-7
Gemini 3.1 Pro Preview Custom ToolsGoogle
92.2-8
Gemini 3.1 Pro PreviewGoogle
92.2-8
GPT-5.4 ProOpenAI
91.9-9
GPT-5.4OpenAI
91.9-10
GPT-5.3 ChatOpenAI
90.5+10
GPT-5.2-CodexOpenAI
90.5-12
GPT-5.2 ChatOpenAI
90.5+40
GPT-5.2 ProOpenAI
90.5-13
GPT-5.2OpenAI
90.5-14
Claude Opus 4.6Anthropic
90.4-15
GPT-5.6 Luna ProOpenAI
89.0-16
GPT-5.6 LunaOpenAI
89.0-17
GPT-5.6 Terra ProOpenAI
89.0-18
GPT-5.6 TerraOpenAI
89.0-19
GPT-5.6 Sol ProOpenAI
89.0-20
GPT-5.6 SolOpenAI
89.0-21
GPT-5.1-Codex-MaxOpenAI
88.7-19
GPT-5.1OpenAI
88.7-19
GPT-5.1-CodexOpenAI
88.7-19
GPT-5 ProOpenAI
88.7-26
GPT-5OpenAI
88.7-28
Gemini 3 Flash PreviewGoogle
88.4-29
GPT-5.1-Codex-MiniOpenAI
87.8-28
DeepSeek V4 ProDeepSeek
86.7-27
o3 ProOpenAI
86.7-27
o3OpenAI
86.7-28
Claude Sonnet 5Anthropic
85.2+123
Claude Sonnet 4.6Anthropic
85.2-30
Claude Opus 4.5Anthropic
85.1-31
Gemini 2.5 ProGoogle
83.5-32
Gemini 2.5 Pro Preview 06-05Google
83.5-33
Gemini 2.5 Pro Preview 05-06Google
83.5-33
Claude Sonnet 4.5Anthropic
82.4-33
Claude Opus 4Anthropic
82.1-35
o4 MiniOpenAI
81.4-36
DeepSeek V3.2DeepSeek
81.3-37
Gemma 4 31BGoogle
80.5-36
Gemma 4 31B (free)Google
80.5-36
Muse Spark 1.1meta
80.2-40
Gemini 3.6 FlashGoogle
80.0-38
Qwen3.5 397B A17BAlibaba
79.4-39
R1 0528DeepSeek
79.4-39
GPT-5.4 NanoOpenAI
79.3-39
GPT-5.4 MiniOpenAI
79.3-40
Gemini 3.1 Flash Lite PreviewGoogle
79.3-41
Gemini 2.5 Flash LiteGoogle
79.1-41
Gemini 2.5 FlashGoogle
79.1-42
Gemini 3.5 FlashGoogle
79.0-43
GLM 5.2Zhipu AI
78.5-44
GLM 5.1Zhipu AI
78.0-32
GLM 5 TurboZhipu AI
78.0-8
MiniMax M2.5MiniMax
78.0-48
GLM 5Zhipu AI
78.0-48
Qwen3.5-122B-A10BAlibaba
77.7-46
Gemma 2 27BGoogle
77.4-46
Qwen3.5-27BAlibaba
77.0-45
Gemini 3.5 Flash LiteGoogle
76.9-45
MiMo-V2.5-ProXiaomi
76.2-47
Qwen3.5-35B-A3BAlibaba
76.0-47
Qwen3.7 PlusAlibaba
75.9-50
Kimi K2.6Moonshot AI
75.8-48
o3 MiniOpenAI
75.3-47
GLM 4.7Zhipu AI
75.1-33
GLM 4.6Zhipu AI
75.1-21
GLM 4.5Zhipu AI
75.1-49
Qwen3.7 MaxAlibaba
74.8+71
Qwen3.5 Plus 2026-04-20Alibaba
74.8+83
Qwen3.6 Max PreviewAlibaba
74.8-51
Claude Sonnet 4Anthropic
74.4-50
o1OpenAI
74.4-50
MiniMax M3MiniMax
74.3-51
R1DeepSeek
74.2-52
Qwen3.6 PlusAlibaba
74.1-55
Inklingthinkingmachines
74.0-54
Hy3Tencent
74.0-61
o1-proOpenAI
73.6-55
MiMo-V2.5Xiaomi
73.0-56
Gemma 4 26B A4B Google
73.0-56
Gemma 4 26B A4B (free)Google
73.0-56
Inkling Smallthinkingmachines
72.4-51
GLM 5V TurboZhipu AI
72.3-57
o4 Mini HighOpenAI
72.1-57
MiniMax M2.1MiniMax
72.0-45
MiniMax M2MiniMax
72.0-58
DeepSeek V3.2 ExpDeepSeek
71.8-49
DeepSeek V3.1DeepSeek
71.8-46
DeepSeek V3 0324DeepSeek
71.8-61
Mistral Medium 3.5Mistral AI
71.7-61
GPT-4o (2024-08-06)OpenAI
71.2-61
GPT-4oOpenAI
71.2-61
GPT-4o (2024-05-13)OpenAI
71.2-61
MiniMax M1MiniMax
70.8-60
GLM 4.5 AirZhipu AI
70.7-60
Claude Haiku 4.5Anthropic
69.5-56
DeepSeek V3DeepSeek
69.5-56
Qwen3 VL 235B A22B ThinkingAlibaba
69.4-47
Qwen3 VL 235B A22B InstructAlibaba
69.4-57
DeepSeek V3.1 TerminusDeepSeek
69.3-57
GPT-4o-miniOpenAI
69.3-57
MiniMax M2-herMiniMax
69.1-56
Qwen3.5-FlashAlibaba
68.6-56
Hy3 previewTencent
68.2-59
Qwen3.5 Plus 2026-02-15Alibaba
68.2+55
Qwen3 Max ThinkingAlibaba
68.2-58
Llama 4 MaverickMeta
67.9-58
GPT-4.1OpenAI
67.7-58
Qwen3 MaxAlibaba
67.4-58
Mistral Large 3 2512Mistral AI
67.0-58
Qwen3 Next 80B A3B ThinkingAlibaba
66.9-41
Qwen3 Next 80B A3B InstructAlibaba
66.9-59
Llama 3.3 70B InstructMeta
66.8-59
GPT-4 TurboOpenAI
66.7-59
Qwen3.5-9BAlibaba
66.5-59
Step 3.5 FlashStepFun
66.1-59
Mistral Large 2407Mistral AI
65.9-33
Mistral LargeMistral AI
65.9-60
Composer 2Cursor
65.7-60
Composer 2 FastCursor
65.7-60
GLM 4.6VZhipu AI
65.5-60
Qwen3 235B A22B Thinking 2507Alibaba
65.5-60
Llama 3.1 70B InstructMeta
65.3-60
GPT-4 Turbo PreviewOpenAI
64.8-45
GPT-4OpenAI
64.8-61
Qwen3 235B A22B Instruct 2507Alibaba
64.7-61
Qwen3 30B A3B Thinking 2507Alibaba
64.1-59
Qwen3 30B A3B Instruct 2507Alibaba
64.1+65
Qwen3 30B A3BAlibaba
64.1-63
o3 Mini HighOpenAI
64.1-63
Trinity Large Thinkingarcee-ai
63.9-62
GPT-5 MiniOpenAI
63.9-61
GLM 4.7 FlashZhipu AI
63.7-61
Mixtral 8x22B InstructMistral AI
63.4-61
GLM 4.5VZhipu AI
62.5-61
Mercury 2Inception
61.0-63
Qwen3 8BAlibaba
61.0-63
Nova 2 LiteAmazon
60.5-63
Phi 4Microsoft
60.2-63
Kimi K2.5Moonshot AI
59.1-62
gpt-oss-20bOpenAI
57.4-64
gpt-oss-20b (free)OpenAI
57.4-64
GPT-4o-mini (2024-07-18)OpenAI
56.5-64
Granite 4.1 8BIBM
55.3-63
Olmo 3 32B ThinkAllen AI
54.9-63
Llama 4 ScoutMeta
54.7-63
Kimi K2.7 CodeMoonshot AI
54.1-63
Qwen3 235B A22BAlibaba
54.0-64
GPT-4.1 MiniOpenAI
53.6-64
Kimi K2 ThinkingMoonshot AI
53.3-64
Kimi K2 0905Moonshot AI
52.4-63
Kimi K2 0711Moonshot AI
51.4-63
Claude 3 HaikuAnthropic
51.3-63
Command ACohere
50.8-63
Command R (08-2024)Cohere
48.7-62
Command R+ (08-2024)Cohere
48.7-63
GPT-5 NanoOpenAI
46.9-63
Llama 3.1 8B InstructMeta
44.5-63
GPT-4.1 NanoOpenAI
42.1-63
R1 Distill Llama 70BDeepSeek
40.7-63
gpt-oss-120bOpenAI
40.4-63
DeepSeek V4 Flash Latest~deepseek
40.0-64
DeepSeek V4 Flash 0731DeepSeek
40.0-64
Qwen3.7 FlashAlibaba
40.0-64
Laguna S 2.1poolside
40.0-63
Laguna S 2.1 (free)poolside
40.0-63
LongCat 2.0Meituan
40.0-63
Kimi K3Moonshot AI
40.0-63
KAT-Coder-Air V2.5Kuaishou
40.0-63
KAT-Coder-Pro V2.5Kuaishou
40.0-63
Grok Latest~x-ai
40.0-63
Aion-3.0-Miniaion-labs
40.0-63
Aion-3.0aion-labs
40.0-63
Laguna XS 2.1poolside
40.0-63
Laguna XS 2.1 (free)poolside
40.0-63
Fugu Ultrasakana
40.0-63
North Mini Code (free)Cohere
40.0-63
Claude Fable Latest~anthropic
40.0-63
Nemotron 3.5 Content Safety (free)NVIDIA
40.0-63
Nemotron 3 UltraNVIDIA
40.0-63
Nemotron 3 Ultra (free)NVIDIA
40.0-64
Step 3.7 FlashStepFun
40.0-64
Perceptron Mk1perceptron
40.0-63
Ring-2.6-1Tinclusionai
40.0-63
GPT Chat LatestOpenAI
40.0-63
Nemotron 3 Nano Omni (free)NVIDIA
40.0-63
Anthropic Claude Haiku Latest~anthropic
40.0-63
OpenAI GPT Mini Latest~openai
40.0-63
Google Gemini Pro Latest~google
40.0-63
MoonshotAI Kimi Latest~moonshotai
40.0-63
Google Gemini Flash Latest~google
40.0-63
Anthropic Claude Sonnet Latest~anthropic
40.0-63
OpenAI GPT Latest~openai
40.0-63
Qwen3.6 FlashAlibaba
40.0-62
Qwen3.6 35B A3BAlibaba
40.0-62
Qwen3.6 27BAlibaba
40.0-62
Ling-2.6-1Tinclusionai
40.0-63
Ling-2.6-flashinclusionai
40.0-63
Claude Opus Latest~anthropic
40.0-63
Lyria 3 Pro PreviewGoogle
40.0-63
Lyria 3 Clip PreviewGoogle
40.0-63
KAT-Coder-Pro V2Kuaishou
40.0-63
Reka Edgerekaai
40.0-63
Mistral Small 4Mistral AI
40.0-63
Nemotron 3 SuperNVIDIA
40.0-63
Nemotron 3 Super (free)NVIDIA
40.0-63
Seed-2.0-LiteByteDance
40.0-63
Seed-2.0-MiniByteDance
40.0-63
Aion-2.0aion-labs
40.0-63
Qwen3 Coder NextAlibaba
40.0-62
Solar Pro 3Upstage
40.0-62
Palmyra X5Writer
40.0-62
GPT AudioOpenAI
40.0-62
GPT Audio MiniOpenAI
40.0-62
Seed 1.6 FlashByteDance
40.0-65
Seed 1.6ByteDance
40.0-65
Nemotron 3 Nano 30B A3BNVIDIA
40.0-65
Nemotron 3 Nano 30B A3B (free)NVIDIA
40.0-65
7

排行榜快照

本周综合评分前10的AI模型。

#模型评分7天变化
1Claude Fable 5Anthropic97.10
2Claude Fable 5 (batch)Anthropic97.1+377
3Claude Opus 5 (Fast)Anthropic95.1+171
4Claude Opus 5Anthropic95.1+171
5Claude Opus 4.8 (Fast)Anthropic95.1-1
6Claude Opus 4.8Anthropic95.1-1
7Claude Opus 4.7 (Fast)Anthropic95.1-5
8Claude Opus 4.7Anthropic95.1-5
9Claude Opus 4.7 (batch)Anthropic95.1+370
10Claude Opus 4.8 (batch)Anthropic94.6+369
8

本周数据

总排名数

300

平均评分

67.8

前10变动

9/10

最活跃的服务商

OpenAI

9

阅读更多分析

性能衰退追踪器
稳定性报告
显著性追踪器
热门模型
Frequently Asked Questions

The weekly report summarizes seven days of AI model ranking changes. It highlights the biggest gainers and losers by rank position, new models entering the leaderboard, models in a fragile watch-list state, and a snapshot of the current top 10. All data comes from hourly-updated composite scores across 290+ models.

Gainers and losers are ranked by their 7-day rank change, which measures how many positions a model moved up or down on the leaderboard over the past week. The models with the largest positive changes are the top gainers, and those with the largest negative changes are the top losers.

The watch list contains models currently in a "fragile" state, meaning their rankings are unstable and could degrade further. Monitoring these models is important if you rely on them in production, as they may experience significant quality or performance drops in the near term.

Weekly AI Performance Report - This Week in AI Models | LM Market Cap