# AVORBYSS Third-Party Notices

**Audit snapshot:** 13 July 2026. This register is an engineering record, not legal advice or final commercial clearance.

AVORBYSS contains original Company-controlled work and depends on third-party components. Third-party licences grant defined permissions; they do not transfer authorship, trade marks, or exclusive ownership to ACN 700 220 784 PTY LTD.

## Bundled browser runtime

### WebLLM

- Package: `@mlc-ai/web-llm`
- Version: `0.2.84`, locked in `pnpm-lock.yaml`
- Project: [mlc-ai/web-llm](https://github.com/mlc-ai/web-llm)
- Declared licence: Apache License 2.0
- Use: compiled into the AVORBYSS Sovereign Device Core and its browser worker
- Licence copy: `THIRD_PARTY_LICENSES/WebLLM-Apache-2.0.txt`

The package's own licence text also identifies third-party protocol material. Before a commercial release, generate and review the complete production dependency SBOM and retain every required notice for WebLLM and its transitive runtime components.

### loglevel 1.9.2

- Package: `loglevel`, resolved as WebLLM's runtime dependency
- Project: [pimterry/loglevel](https://github.com/pimterry/loglevel)
- Declared licence: MIT
- Copyright: Copyright (c) 2013 Tim Perry
- Licence copy: `THIRD_PARTY_LICENSES/loglevel-LICENSE-MIT.txt`

## Runtime-fetched model chain

### Upstream SmolLM2-360M-Instruct

- Publisher: Hugging FaceTB
- Model card: [HuggingFaceTB/SmolLM2-360M-Instruct](https://huggingface.co/HuggingFaceTB/SmolLM2-360M-Instruct)
- Model-card licence declaration: Apache-2.0
- Ownership: third-party open-weight model; not Company-authored

### MLC q4f16 conversion

- Repository: [mlc-ai/SmolLM2-360M-Instruct-q4f16_1-MLC](https://huggingface.co/mlc-ai/SmolLM2-360M-Instruct-q4f16_1-MLC)
- Pinned revision: `3a622fd89e0216e8bb10c410c007c786baa8a033`
- Status: runtime-fetched alpha artifact; commercial provenance and redistribution audit required

### MLC q4f32 conversion

- Repository: [mlc-ai/SmolLM2-360M-Instruct-q4f32_1-MLC](https://huggingface.co/mlc-ai/SmolLM2-360M-Instruct-q4f32_1-MLC)
- Pinned revision: `1e6b2ca3684e2bc8a6d39eb36e8d6cd51eb12bda`
- Status: runtime-fetched alpha artifact; commercial provenance and redistribution audit required

### Binary MLC WebAssembly model libraries

- Repository: [mlc-ai/binary-mlc-llm-libs](https://github.com/mlc-ai/binary-mlc-llm-libs)
- Pinned commit: `025bcaf3780fa8254f5e5efd3bfea0a5397248f4`
- Path series: `web-llm-models/v0_2_84/base`
- Status: runtime-fetched alpha binary; complete licence/build provenance was not established in this audit and is required before commercial release

## Commercial release condition

The current alpha may not be described as containing Company-owned model weights or a completely cleared third-party model chain. Before accepting payment for this model cell, retain written evidence for:

- the base model licence and version;
- the quantisation/conversion source, rights, and modifications;
- the WebAssembly binary source, build recipe, licence, and notices;
- all tokenizer/config files and transitive runtime components;
- allowed commercial use, hosting, caching, mirroring, and redistribution; and
- the production SBOM, vulnerability review, and notice bundle.

If that evidence cannot be obtained, replace the cell or reproducibly compile it from an approved source chain.