Company10 companies
The 10 companies in the ranking
Anthropic publishes the highest-scoring assistant, Claude (74 out of 100), and the best-ranked model on general intelligence, Claude Fable 5.2. OpenAI has the most measured models, 14; Microsoft and Perplexity have none of their own.
AnthropicSan Francisco, California, United States, founded in 2021AssistantClaude · score 74, 1st of 10Measured models12 · best ranked: Claude Fable 5.2, 1st of 71
GoogleMountain View, California, United States, founded in 1998AssistantGemini · score 71, 2nd of 10Measured models11 · best ranked: Gemini 4 Argon, 7th of 71
OpenAISan Francisco, California, United States, founded in 2015AssistantChatGPT · score 62, 3rd of 10Measured models14 · best ranked: GPT-6 Astra, 2nd of 71
MicrosoftRedmond, Washington, United States, founded in 1975AssistantCopilot · score 58, 4th of 10Measured models0 · Copilot serves models from other companies
MetaMenlo Park, California, United States, founded in 2004AssistantMeta AI · score 48, 5th of 10Measured models7 · best ranked: Muse Wonder, 15th of 71
Mistral AIParis, France, founded in 2023AssistantVibe · score 40, 6th of 10Measured models3 · best ranked: Mistral Large 4, 45th of 71
AlibabaHangzhou, China, founded in 1999AssistantQwen · score 39, 7th of 10Measured models13 · best ranked: Qwen3.8 Max, 29th of 71
PerplexitySan Francisco, California, United States, founded in 2022AssistantPerplexity · score 36, 8th of 10Measured models0 · Perplexity serves models from other companies
DeepSeekHangzhou, China, founded in 2023AssistantDeepSeek · score 35, 9th of 10Measured models4 · best ranked: DeepSeek V4.1 Flash 1017, 21st of 71
SpaceX Super IntelligencePalo Alto, California, United States, founded in 2023AssistantGrok · score 26, 10th of 10Measured models4 · best ranked: Grok 4.6, 26th of 71
Scores as of 11/11/2026; models: Epoch AI and LMArena (CC BY 4.0), data as of 10/11/2026. Founding and headquarters: sources on each profile.
Continue in the ranking
The full ranking →Ranking10 AI assistantsRanking of AI assistantsThe Celsius score: one third performance, one third ethics, one third environment.The ranking →RankingAI modelsPerformance of AI models71models ranked on general intelligenceGeneral intelligence, agent tasks, code and the electricity use of a task.The rankings →MethodologyRanking methodology and dataScore composition, criteria and scales, sources and files to download.The methodology →
The hub newsletter
This month's ranking, news on the impact of AI and one figure from the AI Impact Database. One newsletter a month at most, free of charge. Past issues
