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

每周AI性能报告

本周 November 8, 2026

38 个模型上升， 258 个下降， 4 个不变，本周共追踪 300 个模型。

1

执行摘要

最大赢家

Grok 4.7 (batch)

+382 个名次

当前排名 #57

最大输家

Llama 3.1 70B Instruct

-21 个名次

当前排名 #156

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
Grok 4.7 (batch)xAI88.3+382#57
Grok 4.6 (batch)xAI88.3+381#58
GLM 5.3 FlashX LegalZhipu AI69.7+320#119
GLM 5.2 (batch)Zhipu AI68.9+308#131
GLM 4.7Zhipu AI67.1+288#140
GLM 4.6Zhipu AI67.1+288#141
Claude Opus 5Anthropic95.1+264#6
LFM3-1.7BLiquid AI40.0+263#176
LFM3-1.7B (free)Liquid AI40.0+262#177
Ripple 65B Previewricursiveai40.0+261#178
3

涨幅最大

本周排名提升最大的模型。

1
Grok 4.7 (batch)xAI
+382
2
Grok 4.6 (batch)xAI
+381
3
GLM 5.3 FlashX LegalZhipu AI
+320
4
GLM 5.2 (batch)Zhipu AI
+308
5
GLM 4.7Zhipu AI
+288
6
GLM 4.6Zhipu AI
+288
7
Claude Opus 5Anthropic
+264
8
LFM3-1.7BLiquid AI
+263
9
LFM3-1.7B (free)Liquid AI
+262
10
Ripple 65B Previewricursiveai
+261
4

跌幅最大

本周排名下降最大的模型。

1
Llama 3.1 70B InstructMeta
-21
2
MiniMax M2.7MiniMax
-18
3
Inkling Smallthinkingmachines
-18
4
Qwen3.7 FlashAlibaba
-18
5
Nova Lite 1.0Amazon
-18
6
Nova Micro 1.0Amazon
-18
7
Nova Pro 1.0Amazon
-18
8
Muse Spark 1.2meta
-17
9
DeepSeek V4 Flash Latest~deepseek
-17
10
DeepSeek V4 Flash 0731DeepSeek
-17
5

新模型

10 个新模型本周进入排名。

模型评分排名
Grok 4.7 (batch)xAI88.3#57
Grok 4.6 (batch)xAI88.3#58
GLM 5.3 FlashX LegalZhipu AI69.7#119
GLM 5.2 (batch)Zhipu AI68.9#131
LFM3-1.7BLiquid AI40.0#176
LFM3-1.7B (free)Liquid AI40.0#177
Ripple 65B Previewricursiveai40.0#178
Hy4Tencent40.0#179
DeepSeek V4.1 Flash 1017 (batch)DeepSeek40.0#189
Qwen3.8 27B (batch)Alibaba40.0#264
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 5.5Anthropic
95.1+210
Claude Opus 5Anthropic
95.1+264
GPT-5.5 ProOpenAI
92.7+6
GPT-5.3-CodexOpenAI
90.5+17
GPT-5.2-CodexOpenAI
90.5+17
GPT-5.2 ChatOpenAI
90.5+35
Grok 4.7xAI
88.8+11
Grok 4.6xAI
88.8+11
Grok 4.5xAI
88.8+11
Grok 4.20xAI
88.8-9
GPT-5.2 (batch)OpenAI
88.7-7
GPT-5.1-Codex-MaxOpenAI
88.7-7
GPT-5.1OpenAI
88.7+8
GPT-5.1 (batch)OpenAI
88.7-8
GPT-5.1-CodexOpenAI
88.7-8
GPT-5.1-Codex-MiniOpenAI
88.7-8
GPT-5 ProOpenAI
88.7-8
GPT-5 Pro (batch)OpenAI
88.7-8
GPT-5OpenAI
88.7-8
GPT-5 (batch)OpenAI
88.7-8
Grok 4.3 (batch)xAI
88.5-7
Gemini 3 Flash PreviewGoogle
88.4-7
Gemini 3 Flash Preview (batch)Google
88.4-7
Grok 4.20 Multi-AgentxAI
87.9-6
Claude Sonnet 5.5Anthropic
85.2+136
Claude Sonnet 5Anthropic
85.2+221
Claude Sonnet 4.6Anthropic
85.2-6
Claude Opus 4.5Anthropic
85.1-6
Claude Opus 4.5 (batch)Anthropic
85.1-6
Claude Opus 4.8 (batch)Anthropic
83.5-6
Claude Opus 4.7 (batch)Anthropic
83.5-6
Claude Opus 4.6 (batch)Anthropic
83.5-6
Gemini 2.5 ProGoogle
83.5-6
Gemini 2.5 Pro Preview 06-05Google
83.5-6
DeepSeek V3.2DeepSeek
83.4-6
Claude Opus 4.1 (batch)Anthropic
82.9-6
Claude Sonnet 4.5Anthropic
82.4-6
Claude Opus 4.1Anthropic
82.1-6
GPT-6 AstraOpenAI
81.8-6
GPT-6 Astra (batch)OpenAI
81.8-6
GPT-6 Astra ProOpenAI
81.8-6
GPT-6 Astra Pro (batch)OpenAI
81.8-6
o4 Mini (batch)OpenAI
81.4-6
Gemini 3.8 FlashGoogle
81.0-6
Gemini 3.8 Flash (batch)Google
81.0-6
R1 0528DeepSeek
79.4-6
GPT-5.4 NanoOpenAI
79.3-6
GPT-5.4 Nano (batch)OpenAI
79.3-6
GPT-5.4 MiniOpenAI
79.3-6
GPT-5.4 Mini (batch)OpenAI
79.3-6
Gemini 2.5 Flash LiteGoogle
79.1-6
Gemini 2.5 FlashGoogle
79.1-6
Gemini 3.5 Flash LiteGoogle
79.0-6
Gemini 3.5 Flash Lite (batch)Google
79.0-6
Gemini 3.5 FlashGoogle
79.0-6
Gemini 3.5 Flash (batch)Google
79.0-6
Gemma 2 27BGoogle
77.4-6
GPT-5 Mini (batch)OpenAI
76.8-6
GPT-5 NanoOpenAI
76.8-6
GPT-5 Nano (batch)OpenAI
76.8-6
Gemma 4 31BGoogle
75.9-6
Gemma 4 31B (free)Google
75.9-6
Claude Sonnet 4.6 (batch)Anthropic
75.3-6
Claude Sonnet 4.5 (batch)Anthropic
75.3-6
o3 Mini (batch)OpenAI
74.3-6
GPT-4o (batch)OpenAI
72.0-6
DeepSeek V3.2 ExpDeepSeek
71.8+253
DeepSeek V3.1DeepSeek
71.8+258
DeepSeek V3 0324DeepSeek
71.8-8
GPT-4o (2024-11-20)OpenAI
71.2+48
GPT-4o (2024-08-06)OpenAI
71.2-9
GPT-4oOpenAI
71.2-9
GPT-4o (2024-05-13)OpenAI
71.2-9
Gemini 3.1 Flash Lite PreviewGoogle
70.7-9
Qwen3.5 397B A17BAlibaba
70.7-9
Llama 4 MaverickMeta
70.4-9
Claude Haiku 5.5Anthropic
69.9+37
Claude Haiku 4.5Anthropic
69.9-9
Claude Haiku 4.5 (batch)Anthropic
69.9-9
Qwen3.5-122B-A10BAlibaba
69.8-9
GLM 5.3 FlashXZhipu AI
69.7-10
GLM 5.3 FlashZhipu AI
69.7-10
GLM 5.3 Flash (batch)Zhipu AI
69.7-10
DeepSeek V3DeepSeek
69.5-10
Qwen3.5-27BAlibaba
69.3-10
GPT-4o-miniOpenAI
69.3-10
GLM 5V TurboZhipu AI
69.2-10
GLM 5.3 PrimeZhipu AI
68.9-10
GLM 5.3Zhipu AI
68.9-10
GLM 5.3 (batch)Zhipu AI
68.9-10
GLM 5.2Zhipu AI
68.9-10
Qwen3.5-35B-A3BAlibaba
68.7-11
Mistral Large 4Mistral AI
68.3-11
GLM 5.1Zhipu AI
68.3-11
GLM 5 TurboZhipu AI
68.3+10
GLM 5Zhipu AI
68.3-12
GLM 4.5VZhipu AI
67.7-12
GPT-4.1OpenAI
67.7-12
GPT-4.1 (batch)OpenAI
67.7-12
GLM 4.7Zhipu AI
67.1+288
GLM 4.6Zhipu AI
67.1+288
GLM 4.5Zhipu AI
67.1-14
Mistral Large 3 2512Mistral AI
67.0-14
Llama 3.3 70B InstructMeta
66.8-14
Qwen3.5-9BAlibaba
66.5-14
Mistral Large 2407Mistral AI
65.9+253
Mistral LargeMistral AI
65.9-15
Composer 2Cursor
65.7-15
Composer 2 FastCursor
65.7-15
Qwen3 235B A22B Thinking 2507Alibaba
64.7-12
Qwen3 235B A22B Instruct 2507Alibaba
64.7-14
MiniMax M1MiniMax
64.1-12
GLM 4.5 AirZhipu AI
63.7-12
MiniMax M2.7MiniMax
63.6-18
MiniMax M2.5MiniMax
63.6-13
Llama 3.1 70B InstructMeta
63.6-21
Mixtral 8x22B InstructMistral AI
63.4-14
GPT-4o-mini (batch)OpenAI
62.5-14
Phi 4Microsoft
61.0-12
Llama 4 ScoutMeta
60.2-12
GPT-4OpenAI
60.0-12
GPT-4.1 Mini (batch)OpenAI
58.9-12
GPT-4.1 Nano (batch)OpenAI
58.9-12
Claude Haiku 5.5 (batch)Anthropic
58.0-11
GPT-4 Turbo (batch)OpenAI
57.2-11
Command A+Cohere
53.2-9
Command ACohere
51.6-10
Kimi K2.6Moonshot AI
47.5+154
Kimi K2.5Moonshot AI
47.5-11
GPT-4o-mini (2024-07-18)OpenAI
46.3-11
GPT-4.1 MiniOpenAI
44.2-11
Llama 3.1 8B InstructMeta
40.7-11
GPT-5 MiniOpenAI
40.5-11
Command R (08-2024)Cohere
40.5-10
Command R+ (08-2024)Cohere
40.5-11
Particle-1particle-ai
40.0-15
GPT-6.1 Terra ProOpenAI
40.0-15
GPT-6.1 Terra Pro (batch)OpenAI
40.0-15
GPT-6.1 TerraOpenAI
40.0-15
GPT-6.1 Terra (batch)OpenAI
40.0-15
Kimi K3.1 ProMoonshot AI
40.0-15
Flint (free)recursive
40.0-15
MiniMax M3.1 FlashMiniMax
40.0-15
DeepSeek V4.1 Flash 1017DeepSeek
40.0-15
Glimpseelorian
40.0-16
Xing4.0 29B A4B (free)xingchen-agi
40.0-16
Beam 1.0 501B A23Breflection
40.0-16
Ternary Bonsai 2 9Bprism-ml
40.0-15
Moondream 3 Previewmoondream
40.0-15
Moondream 3.1moondream
40.0-15
Muse Spark Openmeta
40.0-15
Nemotron 3.5 Super VLNVIDIA
40.0-15
Nemotron 3.5 Super VL (free)NVIDIA
40.0-15
Pareto 26.10unbiased
40.0-15
Pareto Latest~unbiased
40.0-15
Pareto 26.9unbiased
40.0-15
LFM2.5-24B-A2BLiquid AI
40.0-15
Gemini 4 ArgonGoogle
40.0-15
Gemini 4 Argon (batch)Google
40.0-15
Summarizescaledown
40.0-15
Extractscaledown
40.0-15
Muse Wondermeta
40.0-15
Step 5 PreviewStepFun
40.0-15
Ling 3.1 Flashinclusionai
40.0-15
GPT-6.1 Sol ProOpenAI
40.0-15
GPT-6.1 Sol Pro (batch)OpenAI
40.0-15
GPT-6.1 SolOpenAI
40.0-15
GPT-6.1 Sol (batch)OpenAI
40.0-15
Claude Sonnet 5.5 (batch)Anthropic
40.0-14
Jev Routertypesafe
40.0+190
Perceptron Mk1.5perceptron
40.0-15
Ember-1fireworks
40.0-15
Qwen3.8 Max PrimeAlibaba
40.0-15
Aion 3.5 Miniaion-labs
40.0-15
Aion 3.5aion-labs
40.0-15
Solar Mini 4Upstage
40.0-15
GPT-6 Luna ProOpenAI
40.0-15
GPT-6 Luna Pro (batch)OpenAI
40.0-15
GPT-6 LunaOpenAI
40.0-15
GPT-6 Luna (batch)OpenAI
40.0-15
GPT-6 Sol ProOpenAI
40.0-15
GPT-6 Sol Pro (batch)OpenAI
40.0-15
GPT-6 SolOpenAI
40.0-15
GPT-6 Sol (batch)OpenAI
40.0-15
Claude Opus 5.5 (batch)Anthropic
40.0-14
SwitchyardNVIDIA
40.0+175
MiMo-V2.6-Pro-UltraSpeedXiaomi
40.0-15
MiMo-V2.6-FlashXiaomi
40.0-15
MiMo-V2.6-ProXiaomi
40.0-15
Qwen3.8 Omni FlashAlibaba
40.0-15
Ternary Bonsai 2 27Bprism-ml
40.0-15
DeepSeek Pro Latest~deepseek
40.0-15
DeepSeek Flash Latest~deepseek
40.0-15
Schematron V2 Turboinference-net
40.0-15
Schematron V2 Smallinference-net
40.0-15
GPT Astra Latest~openai
40.0-15
GPT Sol Latest~openai
40.0-15
GPT Terra Latest~openai
40.0-15
GPT Luna Latest~openai
40.0-15
Fugu Ultra v2sakana
40.0-15
Fugu Maxsakana
40.0-15
Ling 3.0 Flash VLinclusionai
40.0-15
DeepSeek V4.1 FlashDeepSeek
40.0-15
DeepSeek V4.1 Flash (batch)DeepSeek
40.0-15
Qwen3.8 Max (0902)Alibaba
40.0-14
Muse Spark 1.3 Contributormeta
40.0-14
Muse Spark 1.3meta
40.0-14
Granite 4.2 8BIBM
40.0-14
Hy4 previewTencent
40.0-14
GLM Flash Latest~z-ai
40.0-14
Qwen3.8 FlashAlibaba
40.0-14
Muse Spark 1.2 Contributormeta
40.0-14
DeepSeek V4 Flash Vision ExpDeepSeek
40.0-14
Hy-MT2-1.8BTencent
40.0-14
Hy-MT2-30B-A3BTencent
40.0-14
GLM Latest~z-ai
40.0-14
Hy-MT2-7BTencent
40.0-14
Qwen3.8 27BAlibaba
40.0-14
Dots3-Note Preview (free)dots-studio
40.0-15
Gemini 3.7 FlashGoogle
40.0-15
Gemini 3.7 Flash (batch)Google
40.0-15
Seed 2.1 TurboByteDance
40.0-15
Qwen3.8 2.4T A95BAlibaba
40.0-15
Seed-2.0-CodeByteDance
40.0-15
DeepSeek V4 Pro 0813DeepSeek
40.0-15
LFM2.5-2.6BLiquid AI
40.0-15
LFM2.5-2.6B (free)Liquid AI
40.0-15
Nemotron 3.5 LightningNVIDIA
40.0-15
Nemotron 3.5 Lightning (free)NVIDIA
40.0-15
Sakana Namazusakana
40.0-15
Solar Pro 4Upstage
40.0-15
Muse Glimmer 30Bmeta
40.0-15
Muse Spark 1.2meta
40.0-17
DeepSeek V4 Flash Latest~deepseek
40.0-17
DeepSeek V4 Flash 0731DeepSeek
40.0-17
Inkling Smallthinkingmachines
40.0-18
Qwen3.7 FlashAlibaba
40.0-18
Claude Opus 5 (batch)Anthropic
40.0-17
Ling 3.0 Flashinclusionai
40.0-17
Gemini 3.6 FlashGoogle
40.0-17
Gemini 3.6 Flash (batch)Google
40.0-17
LongCat 2.0Meituan
40.0-17
Inklingthinkingmachines
40.0-17
Kimi K3Moonshot AI
40.0-16
Kimi K3 (batch)Moonshot AI
40.0-16
Muse Spark 1.1meta
40.0-16
Grok Latest~x-ai
40.0-16
Aion-3.0-Miniaion-labs
40.0-16
Aion-3.0aion-labs
40.0-16
Hy3Tencent
40.0-16
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 5.1Anthropic95.90
2Claude Fable 5.1 (batch)Anthropic95.90
3Claude Fable 5Anthropic95.90
4Claude Fable 5 (batch)Anthropic95.90
5Claude Opus 5.5Anthropic95.1+210
6Claude Opus 5Anthropic95.1+264
7Claude Opus 4.8Anthropic95.1-1
8Claude Opus 4.7Anthropic95.1-3
9GPT-5.5 ProOpenAI92.7+6
10GPT-5.5 Pro (batch)OpenAI92.7-3
8

本周数据

总排名数

300

平均评分

61.7

前10变动

6/10

最活跃的服务商

Zhipu AI

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