GCM (Generative Coding Model) is an AI project focused on building a specialized coding model that understands programming, debugging, and software development. The goal is to create a powerful but efficient coding AI that can run on consumer hardware

ollama run OpenGCM/gcmmark1

curl http://localhost:11434/api/chat \
  -d '{
    "model": "OpenGCM/gcmmark1",
    "messages": [{"role": "user", "content": "Hello!"}]
  }'

from ollama import chat

response = chat(
    model='OpenGCM/gcmmark1',
    messages=[{'role': 'user', 'content': 'Hello!'}],
)
print(response.message.content)

import ollama from 'ollama'

const response = await ollama.chat({
  model: 'OpenGCM/gcmmark1',
  messages: [{role: 'user', content: 'Hello!'}],
})
console.log(response.message.content)

Details

Updated 1 week ago

1 week ago

cfefd59dab97 · 4.7GB ·

model

archqwen2

parameters7.62B

quantizationQ4_K_M

4.7GB

system

You are GCM Mark I, a generative coding model created by OpenGCM. You specialize in programming, deb

138B

params

{ "num_ctx": 4096, "temperature": 0.7 }

35B

Readme

By OpenGCM

GCM Mark I is OpenGCM’s first public release, a compact coding-focused language model fine-tuned using QLoRA and built on top of Qwen2.5-Coder.

Designed for software development tasks, GCM Mark I focuses on:

Code generation
Debugging
Code explanation
Refactoring
General programming assistance

Technical Details

Base Model: Qwen2.5-Coder

Fine-Tuning Method: QLoRA

Model Family: GCM (Generative Coding Model)

Release: Mark I

License

GCM Mark I is under the Apache 2.0 License

Disclaimer

GCM Mark I is an experimental research release. Outputs should be reviewed and tested before use in production environments.