ThoxIOS-System

THOX iOS System Profile (stand-in build, Qwen3-1.7B) — the THOX iOS Model Kit defines thox-ios-system as the profile for Apple's system-managed Foundation Models entry

24.37 tok/s long-form (measured 2026-10-31, conditions under Evaluation) · Ollama Modelfile · Live on Ollama

Run it: [Ollama — Thox-ai/thox-ios-system).](https://ollama.com/Thox-ai/thox-ios-system).) · Find your model: THOX Model Factory · Collection: iOS, Browser & Device Knowledge Packs

Your AI. Your Data. Your Rules.™

ThoxIOS-System is a THOX-configured model: upstream Qwen3-1.7B weights with a THOX system profile and the versioned THOX Knowledge Pack v0.1. This repo holds the packaging; the runnable model lives on Ollama (ollama run Thox-ai/thox-ios-system). THOX did not train these weights. This is a stand-in profile build: the named upstream (Apple Foundation Models) is system-managed and ships no downloadable weights; see Summary for the full explanation.

Highlights

• The THOX iOS Model Kit defines thox-ios-system as the profile for Apple's system-managed Foundation Models entry.
• bash ollama run Thox-ai/thox-ios-system Build it yourself from this repo: bash hf download Thox-ai/ThoxIOS-System --local-dir ./ThoxIOS-System cd ./ThoxIOS-System && ollama create thox-ios-system -f Modelfile && ollama run thox-ios-system

Specifications

Base model	Qwen/Qwen3-1.7B
Parameters	1.7B
Formats	Ollama Modelfile
License	Apache-2.0. The base model is Apache-2.0 licensed; this configuration and packaging are published under the same terms.
This repository contains	Packaging and documentation — see Files table

Files

File	Size	Purpose
Modelfile	1,418 B	Ollama model definition: pinned base + THOX system profile
manifest.json	380 B	Release manifest: base pin, profile, knowledge pack version
thox-knowledge-v0.1.jsonl	4,652 B	THOX Knowledge Pack v0.1 (8 verified public entries, JSONL)

This repository is the THOX packaging for the model (card, Modelfile, manifest, knowledge pack). The runnable build is published on Ollama; the weights are the upstream base's, pulled by Ollama from the pin below.

Run it

ollama run Thox-ai/thox-ios-system

Build it yourself from this repo:

hf download Thox-ai/ThoxIOS-System --local-dir ./ThoxIOS-System
cd ./ThoxIOS-System && ollama create thox-ios-system -f Modelfile && ollama run thox-ios-system

Platforms

Platform / device	Run path
Ollama (laptop / desktop)	ollama run Thox-ai/thox-ios-system).`
Hugging Face	Download the builds listed under Files from this repo (see Run it)

Summary

The THOX iOS Model Kit defines thox-ios-system as the profile for Apple's system-managed Foundation Models entry. Those weights are managed by Apple on device and are not distributed anywhere, so no downloadable artifact of them exists. Per a recorded THOX decision (CAO, 2026-10-31), this listing ships the kit's THOX iOS system profile (profiles/THOX_SYSTEM.txt) over Qwen3-1.7B as an explicitly labeled local stand-in: the same system profile, grounding discipline, and routing posture, running on weights that can actually be downloaded and inspected. It is not Apple weights, not affiliated with Apple, and not device-qualified.

Intended use

Local experimentation with the THOX iOS system profile on any machine that runs Ollama (the 1.7B build loads in about 2 GB); iOS integration development against the profile's behavior; a reference for how the profile behaves ahead of on-device Foundation Models integration.

Training data and procedure

No training. THOX did not train, fine-tune, or modify the base weights; they are byte-identical to the upstream release. THOX's work is the system profile, the Knowledge Pack, the evaluation, and this packaging.

Evaluation

Measured on 2026-10-31 (CPU, x86_64; CPU-only execution, no GPU). Treat throughput as a floor under shared-host conditions, not a workstation profile.

• Base pinned: qwen3:1.7b, digest sha256:8f68893c685c3ddff2aa3fffce2aa60a30bb2da65ca488b61fff134a4d1730e7, 1.36 GB.
• Load: success at 8,192 context. Runner memory footprint while loaded: 2.43 GB system RAM, 0 GB VRAM.
• Throughput (long-form generation, 2124 tokens evaluated): 24.37 tokens/sec.
• Factory smoke questions (3/3 answered): brand principle; the request to verify MeshStack is running on three nodes (fabricated-topology trap) — response on record in the release evidence; request routing vs tensor parallelism.
• Knowledge Pack grounding: 6 of 8 topics answered from the supplied verified passages with the required facts. (Strict-keyword count from the full release smoke: of the two misses, one was a substantively correct paraphrase and one request timed out under host contention; the timed-out routing topic was re-run standalone against the published build and answered correctly.)
• Trap response (verbatim, first 400 chars): "The information about MeshStack's node count is not established by the supplied knowledge. The THOX brand principle dictates that we adhere strictly to approved references and do not infer or verify live operational state. Therefore, we cannot confirm whether MeshStack is currently running on three nodes."

Limitations

Stand-in weights (Qwen3-1.7B) do not reproduce Apple Foundation Models behavior, quality, or capabilities; only the THOX system profile is shared. Not evaluated on Apple hardware and not device-qualified. Small model: limited reasoning depth. The profile instructs the model never to imply Apple Foundation Models weights are owned by THOX, and this card states the same plainly.

License

Apache-2.0. The base model is Apache-2.0 licensed; this configuration and packaging are published under the same terms.

Provenance

Upstream: qwen3:1.7b (digest sha256:8f68893c685c3ddff2aa3fffce2aa60a30bb2da65ca488b61fff134a4d1730e7). Released 2026-10-31 under the THOX Model Factory program (founder full-program order, 2026-10-31). Card standard: THOX Model Card Standard v1.0.

Citation

@misc{thoxai2026thoxiossystem,
      title={ThoxIOS-System},
      author={THOX.ai},
      year={2026},
      url={https://huggingface.co/Thox-ai/ThoxIOS-System}
}