Executespec/ganesh-mini-1.0

VISIONPricing:Input $0.32 / Cached $0.064 / Output $1.6Concurrent Unit Cost:1Model Size:2.3BQuant:BF16Context Size:32kTool Calling:SupportedPublished:Oct 1, 2026License:apache-2.0Architecture:Transformer Open Weights Featherless Exclusive Cold

Ganesh Mini is a 2.3 billion parameter text-only coding fine-tune of Qwen/Qwen3.5-2B, developed by Executespec. It is specifically trained for Go, Java, JavaScript, TypeScript, Python, and Rust, with an intended serving mode for non-thinking text generation using an 8,192-token context. This model is optimized for code generation across multiple programming languages, making it suitable for developers seeking a compact, specialized coding assistant.

Loading preview...

Ganesh Mini Overview

Ganesh Mini is a 2.3 billion parameter language model specifically fine-tuned for code generation across multiple programming languages. It is based on the Qwen/Qwen3.5-2B architecture and is designed for text-only coding tasks. The model supports an 8,192-token context length for efficient code generation.

Key Capabilities

  • Multi-language Code Generation: Trained jointly for Go, Java, JavaScript, TypeScript, Python, and Rust.
  • Compact Size: A 2.3B parameter model, making it suitable for deployment in environments with limited resources.
  • Specialized Fine-tuning: Optimized for coding tasks, distinguishing it from general-purpose LLMs.
  • GGUF Support: Available in F16 and Q8_0 GGUF formats for various deployment scenarios.

Public Assessment & Limitations

A public diagnostic using MultiPL-E HumanEval translations showed Ganesh Mini performing comparably to its upstream Qwen base model across 180 and 952 tasks, with a slight overall difference. It is important to note that this is a local diagnostic, not an official leaderboard result, and does not claim general improvement over the base model. The model may produce incorrect, unsafe, or non-compiling code, and its output should be executed in a resource-limited, network-disabled sandbox. Vision behavior has not been evaluated.

Good for

  • Developers needing a specialized, compact model for generating code in Go, Java, JavaScript, TypeScript, Python, and Rust.
  • Environments requiring efficient, text-only code generation with an 8,192-token context.
  • Use cases where a smaller, fine-tuned coding assistant is preferred over larger, more general-purpose models.

Finetunes
1 models