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

which LLM is the best LLM under 36B NOW?

From a Chatbot Arena battleThe same person sent this prompt to two anonymous models at once. Compare it with Perplexitysonar-pro-high’s answer.
  1. User

    which LLM is the best LLM under 36B NOW?

  2. Perplexity Assistant

    The best LLM under 36 billion parameters as of April 2025 is Command R (35B) by Cohere. Here's why it stands out:

    Key Features

    • 35B parameters with a 128k-token context window, enabling processing of ~100 pages of text[2][3].
    • Top-tier performance in retrieval-augmented generation (RAG), code explanations/rewrites, and multi-step reasoning[1][3].
    • Multilingual support across 10 primary languages[2][7].
    • Cost-effective operation with 5x faster output generation than GPT-4 at 50-75% lower cost per token[9].

    Benchmark Performance

    • Scores 1148 on the Hugging Face Open LLM Leaderboard[4], outperforming similar-sized models like Phi-3 Medium 14B (1123) and Mistral 7B v0.2 (1072)[4].
    • Achieves 88.2% accuracy on MMLU benchmark, surpassing GPT-3.5[9].
    • 71.4% success rate in coding tasks, comparable to larger models[9].

    Use Cases

    • Enterprise-grade conversational AI[2]
    • Complex document analysis[3]
    • Real-time multilingual applications[7]
    • Code generation and debugging[9]

    Alternatives worth considering:

    • Mistral Small 3 (24B): Optimized for low-latency tasks (150 tokens/sec)[7]
    • Phi-3 Medium (14B): Microsoft's cost-effective model with 128k context[3]
    • DeepSeek-V2.5: Strong coding capabilities but requires more resources[6]

    Command R's balance of performance, context handling, and cost efficiency makes it the current leader in the <36B parameter class[2][3][9].

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