Updated June 11, 2026

AI Power Rankings

Independent·Frontier AI lab index
Structural AI Power Index
Weighted by long-horizon inputs — compute, research, talent, data.
Index leader
🇺🇸 Google
9.32
Companies
15
across 3 regions (US 8 · China 6 · Europe 1)
Index range
7.35 → 9.32
structural-weighted
Last updated
June 11
2026
Profile
#CompanyHQScorevs #1Δ vs MayEngineDeliverAccel
01🇺🇸GoogleGemini 3.5 Flash & 3.1 Pro, DeepMind, Vertex/Cloud AI, Search AI, AlphaFold, Veo, WaymoUS9.32-0.229.49.48.9
02🇺🇸OpenAIGPT-5.5 / GPT-5.5 Pro, ChatGPT, Codex, Sora, DALL·EUS9.10-0.22+0.069.19.28.9
03🇺🇸AnthropicClaude Fable 5 & Mythos 5, Claude Code, Claude APIUS8.82-0.50+0.218.88.79.2
04🇺🇸MicrosoftCopilot (Office, GitHub, Windows), MAI in-house models, Azure AI Foundry, PhiUS8.60-0.72+0.068.49.08.7
05🇺🇸MetaLlama 4 family, FAIR / Superintelligence Lab, Meta AI assistant, AI adsUS8.46-0.86+0.038.68.57.8
06🇨🇳AlibabaQwen 3.x family, Model Studio, Alibaba Cloud AI, TongyiCN8.23-1.09+0.158.48.27.8
07🇺🇸AmazonAWS Bedrock, Nova models, Trainium/Inferentia, Alexa+, RufusUS8.16-1.16+0.147.98.58.3
08🇨🇳ByteDanceDoubao models, Seed research, Coze, Dreamina, TikTok AICN8.01-1.31-0.028.28.07.5
09🇺🇸xAIGrok 4.3 (Expert tiers), Grok Build, Colossus supercluster, X integrationUS7.94-1.38+0.378.37.67.2
10🇨🇳DeepSeekDeepSeek V4 (Pro/Flash), R-series reasoning, open weightsCN7.71-1.61+0.277.97.67.2
11🇺🇸AppleApple Intelligence (3rd-gen Foundation Models), Gemini-powered Siri, on-device + Private Cloud ComputeUS7.65-1.67+0.497.28.28.2
12🇨🇳BaiduERNIE 5.0, Ernie Bot, Baidu AI Cloud, Kunlun chipsCN7.63-1.69+0.187.87.67.2
13🇫🇷Mistral AIMistral Large 3, Le Chat, La Plateforme, open-weight modelsFR7.36-1.96+0.067.37.57.4
14🇨🇳Moonshot AIKimi K2.6, Kimi Work agent, Kimi chatbotCN7.36-1.96+0.597.47.47.2
15🇨🇳Zhipu AIGLM-5.1, Z.AI platform (STAR Market listing in progress)CN7.35-1.97+0.747.57.37.0

Leaders by Area

Top 3 per evaluation area

Methodology

Each company is independently scored across ten dimensions by a panel of frontier evaluator models. Scores are averaged, weighted by a structural profile that prizes long-horizon inputs, and published with the per-score justifications intact.

Read the methodology →