Coverage

The AI platforms and models your customers use

Customers use many AI assistants and models, and each can describe your business differently. Modulus tests all relevant models and platforms and stays current as they are updated.

Why it matters

No single AI decides how your business is seen. Customers move between assistants, and the same models also answer inside external apps, search tools, and websites. We define coverage around your customers, test every relevant platform and model, and test again as new versions are released.

AI assistants

The most visited AI assistant websites

Ordered as in Similarweb’s worldwide ranking of AI chatbot and tool websites for September 2026, which is based on estimated web traffic. Use through mobile apps and AI built into other products is not counted. Only general-purpose assistants are listed. Companion and role-play sites, image, design, coding, and other specialist tools, corporate websites, and the older chat.openai.com address are left out.[1]

  1. ChatGPTOpenAI
  2. GeminiGoogle
  3. ClaudeAnthropic
  4. DeepSeekDeepSeek
  5. GrokxAI
  6. PerplexityPerplexity AI
  7. Microsoft CopilotMicrosoft
  8. Qwen ChatAlibaba
  9. Meta AIMeta
  10. KimiMoonshot AI
  11. Duck.aiDuckDuckGo
  12. Microsoft 365 CopilotMicrosoft
  13. ManusManus AI
  14. MuseMeta
  15. GensparkGenspark
  16. LinerLiner
  17. PoeQuora
  18. Le ChatMistral AI

Outside this ranking

Large assistants that Similarweb places outside its AI category or does not rank, so they have no comparable position. Modulus also tests these assistants and includes them in client investigations wherever your customers use them.

  • DoubaoByteDance
  • Tencent YuanbaoTencent
  • Z.ai ChatZ.ai

Models

The model families behind the answers

Ranked by the tokens processed through OpenRouter, a widely used API platform, in the seven days to October 9, 2026, totalled by model family by Modulus. Embedding, image, speech, and classification models, finetunes of another maker’s model, and test models released without a named maker are left out. Use through each maker’s own apps and APIs is not included, so families such as GPT, Claude, and Gemini are understated.[2]

Open weights means the maker has published the model so others can download and run it under its license. Open models such as Llama, DeepSeek, Qwen, GLM, and Kimi can also answer inside products that do not name the model behind them.[3]

  1. DeepSeekDeepSeekOpen weights
  2. GLMZ.ai (Zhipu)Open weights
  3. GPTOpenAI
  4. MiMoXiaomiOpen weights
  5. Hy (Hunyuan)TencentOpen weights
  6. ClaudeAnthropic
  7. NemotronNVIDIAOpen weights
  8. GeminiGoogle
  9. StepStepFunSome open weights
  10. QwenAlibabaSome open weights
  11. MuseMetaSome open weights
  12. KimiMoonshot AIOpen weights
  13. SolarUpstage
  14. MiniMaxMiniMaxOpen weights
  15. LagunaPoolsideOpen weights
  16. GrokxAI
  17. GemmaGoogleOpen weights
  18. LinginclusionAI (Ant Group)Some open weights
  19. gpt-ossOpenAIOpen weights
  20. Dots3Dots StudioOpen weights
  21. InklingThinking Machines LabOpen weights
  22. MistralMistral AISome open weights
  23. ApodexApodexOpen weights
  24. NorthCohereOpen weights
  25. LlamaMetaOpen weights
  26. MercuryInception
  27. NovaAmazon
  28. LFMLiquid AIOpen weights
  29. GraniteIBMOpen weights
  30. SeedByteDance

More model families

Below the top 30 on the same measure. Many are open models that people also download and run themselves.

  • ApertusSwiss AI InitiativeOpen weights
  • CommandCohereOpen weights
  • ERNIEBaiduOpen weights
  • EXAONELG AI ResearchOpen weights
  • FalconTechnology Innovation InstituteOpen weights
  • InternLMShanghai AI LaboratoryOpen weights
  • JambaAI21Open weights
  • LongCatMeituanOpen weights
  • MiniCPMOpenBMBOpen weights
  • Nex-NNex AGIOpen weights
  • OLMoAllen Institute for AIOpen weights
  • Perceptron MkPerceptron
  • PhiMicrosoftOpen weights
  • SmolLMHugging FaceOpen weights
  • SonarPerplexity AI
  • TrinityArcee AIOpen weights
  • Yi01.AIOpen weights

How to read these rankings

The two lists measure different things, so a position on one cannot be compared with a position on the other. Both are snapshots that change from month to month and often from week to week. They reflect web traffic and OpenRouter usage, not overall use, and they say nothing about accuracy. Families within about two percent of each other can swap places from one week to the next.

Open models that people run on their own systems rank differently. By Hugging Face downloads, Qwen leads, followed by Gemma.[3]

Product and company names are trademarks of their respective owners. Their appearance here does not imply any affiliation with or endorsement of Modulus.

Put the problem in expert hands.

Bring us the answer that concerns you, the recommendation that excludes you, or the question your customers are asking.

Request a confidential consultation

Sources

  1. [1]Similarweb • Top AI Chatbots and Tools Websites RankingWorldwide ranking by estimated web traffic • September 2026
  2. [2]OpenRouter • LLM RankingsTokens processed by model, summed by family by Modulus • as of October 9, 2026 • Licensed under CC BY 4.0
  3. [3]Hugging Face • Models HubPublished model weights, licenses, and 30-day downloads • October 10, 2026
Request a confidential consultation