9 2 weeks ago

THOX.ai flagship local assistant. Gemma-4-12B QLoRA fine-tune, Q4_K_M. Pairs with ThoxNova-12B-Agent for tool use.

tools thinking
ollama run Thox-ai/thoxnova-12b-core:Q4_K_M

Applications

Claude Code
Claude Code ollama launch claude --model Thox-ai/thoxnova-12b-core:Q4_K_M
OpenCode
OpenCode ollama launch opencode --model Thox-ai/thoxnova-12b-core:Q4_K_M
Hermes Agent
Hermes Agent ollama launch hermes --model Thox-ai/thoxnova-12b-core:Q4_K_M
OpenClaw
OpenClaw ollama launch openclaw --model Thox-ai/thoxnova-12b-core:Q4_K_M

Models

View all →

Readme

ThoxNova-12B-Core

THOX.ai flagship local assistant. Gemma-4-12B QLoRA fine-tune, Q4_K_M. Pairs with ThoxNova-12B-Agent for tool use.

Tags: tools thinking gemma4 12b q4_k_m

At a glance

  • Base model: google/gemma-4-12b-it
  • Fine-tune: QLoRA supervised fine-tune (THOX thoxllm-factory pipeline)
  • Architecture: gemma4
  • Parameters: 11.9B
  • Quantization: Q4_K_M (7.4 GB)
  • Native context: 262144 tokens (default runtime num_ctx 8192)
  • Prompt template: Gemma (<start_of_turn> / <end_of_turn>)
  • License: Gemma Terms of Use — https://ai.google.dev/gemma/terms. 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/thoxnova-12b-core:Q4_K_M   # Claude Code
ollama launch chatgpt  --model thox-ai/thoxnova-12b-core:Q4_K_M   # Codex App
ollama launch openclaw --model thox-ai/thoxnova-12b-core:Q4_K_M   # OpenClaw
ollama launch hermes   --model thox-ai/thoxnova-12b-core:Q4_K_M   # Hermes
ollama launch codex    --model thox-ai/thoxnova-12b-core:Q4_K_M   # Codex
ollama launch opencode --model thox-ai/thoxnova-12b-core:Q4_K_M   # OpenCode

Quick start

ollama pull thox-ai/thoxnova-12b-core:Q4_K_M
ollama run thox-ai/thoxnova-12b-core:Q4_K_M "your prompt"

Intended use

General-purpose local assistant for THOX operator workflows: drafting, summarization, code review, reasoning. Supports thinking.

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

Legal or medical advice; unattended long-horizon autonomy.

Notes

  • Context: the GGUF supports up to 262144 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.