Thox-ai/ thoxforge-7b:Q4_K_M

27 1 month ago

THOX.ai coding and reasoning assistant. Mistral-7B-v0.3 QLoRA fine-tune, Q4_K_M, tuned for Rust / TypeScript / Python. 

ollama run Thox-ai/thoxforge-7b:Q4_K_M

Details

1 month ago

ea2c8b74a636 · 4.4GB ·

model
llama
·
7.25B
·
Q4_K_M
license
Base model: Mistral-7B-Instruct-v0.3 (Apache-2.0). THOX LoRA delta and packaging: Apache-2.0. Publis
license
Base model governed by its own license: Apache-2.0. THOX fine-tune, Modelfile, and packaging: (c) 20
template
{{ if .System }}[INST] <<SYS>> {{ .System }} <</SYS>> {{ .Prompt }} [/INST]{{ else }}[INST] {{ .Prom
system
You are ThoxForge-7B, the THOX.ai coding and reasoning assistant. Answer in clean technical prose. P
params
{ "num_ctx": 8192, "stop": [ "</s>", "[INST]", "[/INST]" ],

Readme

ThoxForge-7B

THOX.ai coding and reasoning assistant. Mistral-7B-v0.3 QLoRA fine-tune, Q4_K_M, tuned for Rust / TypeScript / Python.

Tags: code reasoning mistral 7b q4_k_m

At a glance

  • Base model: mistralai/Mistral-7B-Instruct-v0.3
  • Fine-tune: QLoRA supervised fine-tune (THOX thoxllm-factory pipeline)
  • Architecture: llama (Mistral lineage)
  • Parameters: 7.2B
  • Quantization: Q4_K_M (4.4 GB)
  • Native context: 32768 tokens (default runtime num_ctx 8192)
  • Prompt template: Mistral ([INST] ... [/INST])
  • License: Mistral-7B-Instruct-v0.3 base: Apache-2.0. THOX fine-tune, Modelfile, and packaging: © 2026 THOX.ai LLC — see NOTICE.md.
  • Publisher: THOX.AI LLC

Applications

Launch this model directly into a supported tool:

ollama launch claude   --model thox-ai/thoxforge-7b:Q4_K_M   # Claude Code
ollama launch chatgpt  --model thox-ai/thoxforge-7b:Q4_K_M   # Codex App
ollama launch openclaw --model thox-ai/thoxforge-7b:Q4_K_M   # OpenClaw
ollama launch hermes   --model thox-ai/thoxforge-7b:Q4_K_M   # Hermes
ollama launch codex    --model thox-ai/thoxforge-7b:Q4_K_M   # Codex
ollama launch opencode --model thox-ai/thoxforge-7b:Q4_K_M   # OpenCode

Quick start

ollama pull thox-ai/thoxforge-7b:Q4_K_M
ollama run thox-ai/thoxforge-7b:Q4_K_M "your prompt"

Intended use

Code generation and review with a bias toward Rust, TypeScript, and Python. Produces working code with types and error handling.

The model ships with the THOX factuality system prompt: it answers company questions only from verified facts and declines to invent non-public details (headquarters, headcount, funding, founding date).

Not intended for

Unreviewed autonomous coding; legal or medical advice.

Notes

  • Context: the GGUF supports up to 32768 tokens; the published default num_ctx is 8192 for predictable memory on edge devices. Override at runtime with --num-ctx or a client parameter.
  • On-device: built for local execution on THOX hardware and desktops. Your AI. Your Data. Your Rules.

Links

Published by THOX.AI LLC from the thoxllm-factory pipeline.