cmsmanhattan/ jirack-ultra-7b-q3

1 Download Updated 1 month ago

ollama run cmsmanhattan/jirack-ultra-7b-q3

curl http://localhost:11434/api/chat \
  -d '{
    "model": "cmsmanhattan/jirack-ultra-7b-q3",
    "messages": [{"role": "user", "content": "Hello!"}]
  }'

from ollama import chat

response = chat(
    model='cmsmanhattan/jirack-ultra-7b-q3',
    messages=[{'role': 'user', 'content': 'Hello!'}],
)
print(response.message.content)

import ollama from 'ollama'

const response = await ollama.chat({
  model: 'cmsmanhattan/jirack-ultra-7b-q3',
  messages: [{role: 'user', content: 'Hello!'}],
})
console.log(response.message.content)

Models

View all →

Name

1 model

Size / Usage

Context

Input

jirack-ultra-7b-q3:latest

3.8GB · 128K context window · Text · 1 month ago

jirack-ultra-7b-q3:latest

3.8GB

128K

Text

Readme

JiRack Ultra 7B Q3_K_M

Compact ~7B model from CMS Manhattan. BitNet-style ternary path, extended tokenizer with Routing, Tool-call and Robotics tags. Production Q3_K_M build for tighter memory budgets.

HF: https://huggingface.co/CMSManhattan/JiRackUltra_7b

Run

ollama run cmsmanhattan/jirack-ultra-7b-q3

What it is

Solid step up from 1B for coding, tool-call, and general tasks
Same JiRack tokenizer tags: routing, tools, robotics
Q3_K_M quant for lower VRAM / RAM than Q4
Fits RAG, agents, and Spring / Java tool-call stacks
American ternary stack from CMS Manhattan (New York)

Hardware

Q3_K_M: roughly 3.5–4 GB on disk.
Recommend 5–6 GB VRAM (GPU) or 10–12 GB system RAM (CPU).

Contact

grabko@cmsmanhattan.com
+1 (516) 777-0945
New York, USA

License

MIT on weights unless stated otherwise on the HF card. Commercial terms for UI / enterprise on request.