katiyardinesh/gemma-4-31B-it-custom

VISIONPricing:Input $0.48 / Cached $0.1 / Output $1.44Concurrent Unit Cost:2Model Size:31BQuant:FP8Context Size:32kTool Calling:SupportedPublished:Sep 25, 2026License:apache-2.0Architecture:Transformer Open Weights Featherless Exclusive Cold

The katiyardinesh/gemma-4-31B-it-custom model is a 31 billion parameter instruction-tuned variant from the Gemma 4 family, developed by Google DeepMind. This multimodal model processes text and image inputs, generating text outputs, and features a 32768 token context window. It is designed for advanced reasoning, coding, and agentic workflows, offering strong performance across various benchmarks.

Loading preview...

Overview

Gemma 4 is a family of open, multimodal models developed by Google DeepMind, designed for diverse deployment scenarios from mobile devices to servers. The katiyardinesh/gemma-4-31B-it-custom is a 31 billion parameter instruction-tuned variant, part of the larger Gemma 4 release which includes both Dense and Mixture-of-Experts (MoE) architectures. These models are capable of handling text and image inputs, with some variants also supporting audio and video, and generating text outputs. They feature an extended context window of up to 256K tokens and multilingual support for over 140 languages.

Key Capabilities

  • Multimodality: Processes text, image (with variable aspect ratio and resolution), and video inputs. The E2B, E4B, and 12B models also natively support audio.
  • Reasoning: Designed with configurable thinking modes for step-by-step problem-solving.
  • Extended Context: Supports context windows up to 256K tokens, enabling complex, long-context tasks.
  • Enhanced Coding & Agentic Capabilities: Achieves significant improvements in coding benchmarks and includes native function-calling support for autonomous agents.
  • Native System Prompt Support: Facilitates more structured and controllable conversations.
  • Optimized Architectures: Offers both Dense and MoE variants, with smaller models optimized for on-device execution.

Good For

  • Advanced Reasoning Tasks: Excels in complex problem-solving and logical deduction.
  • Agentic Workflows: Ideal for applications requiring structured tool use and autonomous agent development.
  • Coding: Strong performance in code generation, completion, and correction.
  • Multimodal Understanding: Capable of interpreting and generating responses based on interleaved text, image, and video inputs.
  • Long-Context Applications: Suitable for tasks requiring processing and understanding extensive documents or conversations.

Finetunes
1 models