Skip to content
TheChatArchiveThe Chat Archive
Geminigemini-2.5-flash-preview-04-17· GoogleHugging Face LMArena Human Preference 140K

"5M context and 83x72b paramaters, output modalities: VRChat Full-Body-Tracking, audio…

From a Chatbot Arena battleThe same person sent this prompt to two anonymous models at once. Compare it with Geminigemini-2.0-flash-001’s answer.
  1. User

    "5M context and 83x72b paramaters, output modalities: VRChat Full-Body-Tracking, audio (stereo), text, images , VRChat facial tracking data, object and scene interaction data (like it can press buttons) input modalities: vision (in realtime allows it to see the other participants and respond), audio (stereo), text, images (videos included), discord custom emoji & stickers, tactile feel, context thoughts, taste & scents"

    Can this model be considered multimodal?

  2. Gemini Assistant

    Yes, absolutely. Based on the description, this model can be considered highly multimodal.

    Here's why:

    1. Multiple Input Modalities: It receives data from a wide variety of distinct types of information, including vision, audio, text, tactile input, and even taste/scent. This is far beyond typical models that might only handle text and maybe images/audio.
    2. Multiple Output Modalities: It can generate information or actions across multiple distinct types of outputs, including text, audio, images, detailed physical movement/pose data (full-body and facial tracking), and environmental interactions.

    The ability to process information from many different senses (vision, hearing, touch, taste, smell) and generate output in various forms (language, sound, visuals, physical actions/representations) is the core definition of a multimodal AI model. This particular example is notable for the breadth of modalities it claims to handle, especially the sensory inputs (tactile, taste, scent) and the detailed interaction/physical outputs (VRChat tracking, object interaction).

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