Skip to content
TheChatArchiveThe Chat Archive
gpt-4.1-mini-2025-04-14· OpenAIHugging Face LMArena Human Preference 140K

There's this model that I keep running into when using lmarena. Its called…

From a Chatbot Arena battleThe same person sent this prompt to two anonymous models at once. Compare it with Claudeclaude-opus-4-20250514’s answer.
  1. User

    There's this model that I keep running into when using lmarena. Its called "sonar-reasoning-pro-high", and to me its one of the best models on that site. Where can I find it to use it?

  2. Assistant

    The model "sonar-reasoning-pro-high" you’re referring to on LM Arena is likely a specialized or proprietary model hosted on their platform. LM Arena often features models that are either custom fine-tuned, experimental, or provided by third parties, and not all of them are publicly available for direct download or use outside their environment.

    Where to find or use "sonar-reasoning-pro-high":

    1. On LM Arena itself:
      The primary way to use this model is through the LM Arena platform. They provide an interface where you can run prompts against it directly. If you want to use it, the best approach is to continue using it there.

    2. Check LM Arena documentation or community:
      Sometimes, LM Arena or the model creators share links to the model weights or APIs. Look for any official announcements, GitHub repos, or forums linked from LM Arena’s site.

    3. Contact LM Arena or the model authors:
      If you want to use the model outside LM Arena, you might try reaching out to LM Arena support or the model’s developers (if their info is available) to ask if the model is available for download or via an API.

    4. Look for similar open-source models:
      If the model is proprietary or not publicly released, you might find similar reasoning-focused models on platforms like Hugging Face Hub, or repositories related to reasoning or logic-enhanced LLMs.


    Summary:

    • "sonar-reasoning-pro-high" is likely only accessible on LM Arena’s platform.
    • No public download or API is generally available unless explicitly shared.
    • Use it on LM Arena or inquire with their team for more access options.

    If you want, I can help you find similar open-source models for reasoning tasks! Would you like that?

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