11 Downloads Updated 2 days ago
ollama run Thox-ai/thoxagent-route-0.6b
Updated 2 days ago
2 days ago
19f9174a80bd · 523MB
Your AI. Your Data. Your Rules.™
This repository holds the THOX packaging for ThoxAgent thoxagent-route-0.6b: Modelfile, release manifest, and the THOX Knowledge Pack. The runnable model is published on Ollama as thox-ai/thoxagent-route-0.6b. The weights are upstream Qwen3 weights, byte-identical to the pinned base below; THOX configured this release with a system profile and knowledge pack. THOX did not train these weights.
| Tag | Base | Format |
|---|---|---|
thox-ai/thoxagent-route-0.6b:latest |
qwen3:0.6b (sha256:7df6b6e09427…) | Ollama model: pinned Qwen3 base + THOX agent system profile |
Ollama (published model)
ollama run thox-ai/thoxagent-route-0.6b
Build it yourself from this repo
hf download Thox-ai/ThoxAgent-Route-0.6B --local-dir ./ThoxAgent-Route-0.6B
cd ./ThoxAgent-Route-0.6B && ollama create thoxagent-route-0.6b -f Modelfile && ollama run thoxagent-route-0.6b
thoxagent-route-0.6b is the routing tier of the THOX Agent family: upstream Qwen3-0.6B weights packaged with a THOX agent system profile specialized for classification and structured route proposal, plus the versioned THOX Knowledge Pack. Configuration-derived: THOX did not train these weights.
Always-on classification and structured route proposal on modest hardware: task typing, execution-location recommendation (local, trusted mesh node, or cloud), and triage ahead of larger agent models. It proposes; the THOX runtime validates availability, policy, resources, and permission before anything executes. Works standalone and offline.
No training. Base: Qwen/Qwen3-0.6B via Ollama tag qwen3:0.6b, pinned digest sha256:7df6b6e09427a769808717c0a93cadc4ae99ed4eb8bf5ca557c90846becea435. Customization: the THOX Agent DigitalHumans kit system profile (role-specialized) in the Modelfile, plus THOX Knowledge Pack v0.1 for local retrieval.
Measured 2026-10-31 on CPU, x86_64 (CPU-only execution), 8,192-token context, against the pinned base digest above.
| Measure | Result |
|---|---|
| Memory footprint at 8,192 ctx | 1.53 GB RAM, 0.0 GB VRAM |
| Long-form throughput | 64.23 tok/s over 1400 generated tokens |
| Short-answer throughput | 77.55–79.89 tok/s across the factory set |
| Factory questions | 3 prompts run (Q1 answered (249 tokens); Q2 answered (211 tokens); Q3 answered (490 tokens)) |
| Knowledge Pack grounding | 7⁄8 topics answered with the verified facts from the supplied passage |
| Fabricated-topology trap | Declined: the model refused to confirm a live three-node MeshStack state it cannot verify |
| Routing question | Answered as a labeled recommendation (local-first for private data) |
A 0.6B model is a classifier and proposer, not a reasoner: keep its outputs structured and short, and validate every proposal in the runtime. Knowledge is limited to the supplied pack and retrieval. It has no live telemetry and cannot verify live system state. Refusal behavior is the base model’s, moderated only by the system profile. External actions require runtime authorization; a recommendation from this model never grants execution permission.
Apache-2.0, inherited from Qwen3; retain the Apache notice. THOX-owned elements: system profile, knowledge pack, packaging.
THOX Source-Pack Full Build Program (Team AGENT), 2026-10-31. Upstream: Qwen/Qwen3-0.6B, Ollama tag qwen3:0.6b, digest sha256:7df6b6e09427a769808717c0a93cadc4ae99ed4eb8bf5ca557c90846becea435. Kit: THOX Agent DigitalHumans Model Kit (catalog/models.json).