Skip to content
The Chat Archive
kimi-k2-0711-preview LMArena Human Preference 140K

provide an overview of current marketshares amongst the most popular llm's

  1. User

    provide an overview of current marketshares amongst the most popular llm's

  2. Assistant

    Current LLM Market Share Overview (as of late 2024)

    The LLM market is rapidly evolving, with no single authoritative source for market share. However, based on usage data, API calls, and industry reports, here's a general breakdown:

    1. OpenAI (GPT-4/4o, GPT-3.5)

    • Estimated Share: ~50-60% of paid API market
    • Key Strengths: First-mover advantage, strong developer ecosystem, ChatGPT's 180M+ weekly users
    • Notable: GPT-4o mini rapidly gaining traction for cost-sensitive applications

    2. Anthropic (Claude 3.5 Sonnet/Haiku)

    • Estimated Share: ~15-20% (fastest growing)
    • Key Strengths: Superior reasoning for coding/analysis, 200K context window, strong safety reputation
    • Growth Driver: Increasing enterprise adoption (particularly in finance/legal)

    3. Google (Gemini 1.5 Pro/Flash)

    • Estimated Share: ~10-15%
    • Key Strengths: Native multimodal capabilities, 2M token context, integration with Google Cloud
    • Challenge: Playing catch-up after Bard's rocky launch

    4. Meta (Llama 3.1 405B/70B)

    • Estimated Share: ~5-10% (but dominates open-source)
    • Key Strengths: Free for commercial use, customizable, popular for on-premise deployments
    • Unique Position: Most downloaded open weights model (Llama 3.1 variants)

    5. Cohere (Command R+)

    • Estimated Share: ~3-5%
    • Focus: Enterprise RAG applications, multilingual support (23 languages)

    6. Mistral (Large 2, 8x22B)

    • Estimated Share: ~2-4%
    • Strengths: European alternative, strong performance/parameter ratio

    Key Trends:

    • Open-source surge: Llama 3.1 variants now power >60% of self-hosted LLMs
    • Price wars: GPT-4o mini and Gemini Flash driving 50%+ cost reductions
    • Context competition: Gemini 1.5's 2M tokens vs Claude's 200K vs GPT-4o's 128K
    • Multimodal shift: ~30% of new applications now use vision capabilities

    Note: These figures exclude China (where Baidu's ERNIE and Alibaba's Tongyi dominate) and represent Western markets primarily. The landscape changes monthly with new model releases.

Responses were produced in different, uncontrolled contexts (system prompts, settings, prior turns). This is not a benchmark.