Skip to content
TheChatArchiveThe Chat Archive
Perplexitysonar-pro-high· PerplexityHugging Face LMArena Search Arena 24K

what is the top ten model in livebench now?

From a Chatbot Arena battleThe same person sent this prompt to two anonymous models at once. Compare it with gpt-4o-search-preview’s answer.
  1. User

    what is the top ten model in livebench now?

  2. Perplexity Assistant

    Based on the latest LiveBench results available, here are the current top performing AI models:

    1. Claude 3.7 Sonnet: This model from Anthropic has taken the top spot, excelling in hybrid reasoning and demonstrating exceptional all-around performance across various benchmarks[8].

    2. OpenAI o3-mini (high): This model has shown outstanding performance, particularly in coding tasks. Its coding average on LiveBench is 82.74, significantly higher than other models[10].

    3. OpenAI o1: While slightly behind o3-mini in coding, it still performs strongly across multiple categories[10].

    4. GPT-4o: This cost-optimized variant of GPT-4 from OpenAI remains highly competitive, especially in areas like reasoning and language processing[6][9].

    5. DeepSeek R1: This model has shown strong performance in complex data analysis and logical reasoning[9].

    6. Gemini 2.0 Pro: Google's latest model excels in contextual understanding and large-scale project management[5][9].

    7. Claude 3.5 Sonnet: While surpassed by its successor, this model still performs well across various benchmarks[5][10].

    8. Grok 3: xAI's latest model shows strong performance in advanced reasoning, mathematics, and coding[9].

    9. Llama 3.3 70B: Meta's model demonstrates competitive performance, especially in multilingual tasks[5][9].

    10. Mistral Large 2: This model shows strong performance, particularly in multilingual applications and document synthesis[9].

    It's important to note that the AI landscape is rapidly evolving, and rankings can change quickly as new models are released and existing ones are updated. Additionally, different benchmarks may yield slightly different rankings depending on the specific tasks and metrics used.

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