TaimoorSiddiqui/HopCoder-Mini-35B-A3B-VL36

TEXT GENERATIONConcurrency Cost:3Model Size:35.1BQuant:FP8Ctx Length:32kTool Calling:SupportedLicense:apache-2.0Architecture:Transformer Open Weights Cold

HopCoder-Mini-35B-A3B-VL36 by Taimoor Siddiqui is a 35.1 billion parameter BF16 vision-language model, combining a 35B-A3B agentic text backbone with a VL36 vision stack. This model is designed for agentic coding, tool-calling workflows, and multimodal understanding, supporting both image and video inputs. It excels in long-context chat and coding use cases, offering full-precision weights for robust performance.

Loading preview...

HopCoder Mini-35B-A3B-VL36 Overview

HopCoder Mini-35B-A3B-VL36, developed by Taimoor Siddiqui, is a 35.1 billion parameter vision-language model. It integrates a 35B-A3B agentic text backbone with a VL36 vision stack, enabling comprehensive image and video understanding. The model is released with BF16 full-precision weights and supports a 32768 token context length.

Key Capabilities

  • Multimodal Input: Processes both image and video data through its VL36 vision stack.
  • Agentic Workflows: Designed for agentic coding and tool-calling applications.
  • Long-Context Support: Capable of handling long-context chat and coding scenarios.
  • Full Precision: Utilizes BF16 full-precision weights, ensuring high fidelity in computations.
  • Identity Management: Includes a default chat identity via chat_template.jinja to identify the assistant as HopCoder Mini by Taimoor Siddiqui.

Good For

  • Developers requiring a multimodal model for coding and agentic tasks.
  • Applications that benefit from integrated vision and language capabilities.
  • Use cases demanding long-context processing for complex interactions.

Finetunes
2 models