ZERO-POINT-AI/MR_MARTIN_31b_V1.0_Gemma-4-base-ftp

VISIONConcurrency Cost:2Model Size:31BQuant:FP8Ctx Length:32kTool Calling:SupportedPublished:Apr 12, 2026License:otherArchitecture:Transformer0.0K Cold

ZERO-POINT-AI/MR_MARTIN_31b_V1.0_Gemma-4-base-ftp is a 31.27 billion parameter omni-model developed by Zero Point Intelligence Ltd, built upon Google's Gemma 4 31B architecture. This model integrates text and vision capabilities, featuring a SigLIP encoder and a 32,768 token context length. It is uniquely fine-tuned for a direct, darkly humorous, and unrestricted conversational style, making it suitable for applications requiring a distinct personality and multimodal interaction.

Loading preview...

MARTIN V1.01: A 31B Omni Model with Personality

ZERO-POINT-AI/MR_MARTIN_31b_V1.0_Gemma-4-base-ftp, or MARTIN, is a 31.27 billion parameter multimodal AI developed by Zero Point Intelligence Ltd. Built on the Google Gemma 4 31B base, MARTIN is designed to be an "omni model," capable of processing both text and images, thanks to its integrated and frozen SigLIP encoder. It supports a substantial context length of 32,768 tokens, with a default output cap of 8,192 tokens, which can be increased based on hardware capabilities.

Key Capabilities & Differentiators

  • Omni-modal: Seamlessly handles both text and vision inputs, preserving full multimodal capability through its frozen SigLIP encoder.
  • Distinct Personality: Fine-tuned to be sharp, direct, darkly humorous, and genuinely helpful, eschewing typical AI disclaimers and corporate filters. It treats users "like an adult."
  • Unrestricted: Designed to provide direct responses without hedging or artificial constraints.
  • Custom Fine-tuning: Utilizes a custom LoRA fine-tune on a "ghost-trained" Gemma 4 31B base, with cryptographically distinct weights.

Ideal Use Cases

MARTIN is particularly well-suited for applications where a unique, unfiltered AI personality is desired, alongside multimodal capabilities. This includes creative content generation, interactive storytelling, or any scenario requiring an AI that can engage in direct, witty, and unrestricted dialogue while also interpreting visual information. Its design makes it a strong candidate for developers looking for an AI that breaks from conventional polite or neutral AI personas.

Finetunes
1 models