ThoxFlux-350M

ThoxClip Edge Assistant — an on-device assistant model for the ThoxClip tier: a LoRA fine-tune (rank 32) on a THOX-authored corpus of 982 chat, tool-use, and factuality examples, merged into the base weights

GATE49 ²⁰⁄₄₉ (development set) · Safetensors weights · GGUF: Q4_K_M · Live on Ollama

Your AI. Your Data. Your Rules.™

Status: PUBLISHED — GROUNDING REQUIRED. Deployed surfaces supply facts through the serving layer; the bare weights are not a reliable source of THOX facts. The full gate result is printed under Evaluation.

Summary

ThoxFlux-350M is an on-device assistant model for the ThoxClip tier: a LoRA fine-tune (rank 32) on a THOX-authored corpus of 982 chat, tool-use, and factuality examples, merged into the base weights. Status: PUBLISHED, grounding required. Deployed surfaces supply facts through the serving layer; the bare weights are not a reliable source of THOX facts.

Highlights

• THOX fine-tune: a LoRA fine-tune (rank 32) on a THOX-authored corpus of 982 chat, tool-use, and factuality examples, merged into the base weights.
• GATE49 ²⁰⁄₄₉ on the THOX factuality gate (development set, abstention-weighted), mean 2.⁴⁹⁄₅, 18 hard failure(s) — printed in full under Evaluation.
• This repo carries the merged weights (safetensors), a Q4_K_M GGUF, and an Ollama Modelfile.

Specifications

Files

Run it

This repo carries the merged weights (safetensors), a Q4_K_M GGUF, and an Ollama Modelfile.

ollama pull Thox-ai/thoxflux-350m
ollama run Thox-ai/thoxflux-350m

hf download Thox-ai/ThoxFlux-350M ThoxFlux-350M.Q4_K_M.gguf --local-dir ./ThoxFlux-350M

Platforms

Intended use

• On-device assistant and tool routing for ThoxClip: short conversational replies, THOX house voice, and routing intents to device and mesh functions, fully offline, with facts supplied by the serving layer.
• Out of scope: safety-critical, medical, or legal decisions without a human in the loop, and any use as a standalone source of THOX facts.

Training data and procedure

THOX fine-tune of LiquidAI/LFM2.5-350M. The corpus and procedure are described in this card as published at release (see Summary and Evaluation); no training beyond what this card describes is claimed for this repository.

Evaluation

Measured with the THOX GATE49 factuality gate (49 questions, weights only, no serving-side grounding; GATE49 is a development set, abstention-weighted). Full narrative under Evaluation below.

THOX GATE49 factuality gate (49 questions, weights only, no serving-side grounding), most recent evaluation of the shipped build: ²⁰⁄₄₉ passed, mean 2.⁴⁹⁄₅, 18 hard failures. This is a weak gate result. The build is published for its identity, voice, and tool-routing behaviour under grounding, not for factual recall from the weights.

Limitations

• Factual answers from the bare weights are unreliable; deploy with serving-side grounding for facts, and escalate questions beyond the trained facts through MeshStack.
• A 350M model: small context and a low reasoning ceiling by design. The fine-tune teaches THOX identity, house voice, and tool-routing shapes, not new capability.
• This build has not been qualified on ThoxClip device hardware.
• The license revenue ceiling (below) is a hard constraint on commercial use.

License

LFM Open License v1.0 (see LICENSE in this repo). It is not Apache-2.0. The license restricts commercial use to entities under $10M annual revenue; read the license before any commercial deployment. The THOX fine-tune and additions are covered under the same terms.

THOX.ai — Your AI. Your Data. Your Rules.™

Provenance

Published by THOX.ai as Thox-ai/ThoxFlux-350M. Upstream base: LiquidAI/LFM2.5-350M. This card was conformed to THOX Model Card Standard v1.0 on 2026-10-31 (frontmatter, tags, header, files, and section structure); the technical content is as published at release.

Citation

@misc{thoxai2026thoxflux350m,
      title={ThoxFlux-350M},
      author={THOX.ai},
      year={2026},
      url={https://huggingface.co/Thox-ai/ThoxFlux-350M}
}