Update README.md

README.md

@@ -5,53 +5,79 @@ license_link: LICENSE
 tags:
   - runtime
   - llama.cpp
+  - llama-server
+  - stable-diffusion.cpp
   - tts
+  - text-to-image
+  - image-generation
+  - gguf
+  - on-device
   - openbloom
 ---
 
 # OpenBloom Runtime Releases
 
-This repository hosts **OpenBloom's self-built runtime artifacts** —
-the engines and components that the OpenBloom desktop application
-provisions at install time. It contains **no models**: it hosts the
-*runtime tooling* that serves open-weight models locally, hash-pinned
-and verified on every download.
+This repository hosts **OpenBloom's self-built runtime artifacts** — the engines, bundles and
+converted weights that the OpenBloom desktop application provisions at install time. Every
+artifact is hash-pinned by the OpenBloom downloader and verified fail-closed on every download.
 
-## KrisKross 1.0.0
+**This repository contains no third-party model weights** — with one exception: the MOSS-TTS-Realtime
+GGUF conversion, which is OpenBloom's own quantization of the Apache-2.0 MOSS-TTS model.
 
-The KrisKross runtime bundle: OpenBloom's build of the MOSS-TTS
-inference runtime.
+## Contents
 
-| Component | Source | License |
+| Family | Path | What it is |
 |---|---|---|
-| `llama-moss-tts` | Built from [OpenMOSS/llama.cpp](https://github.com/OpenMOSS/llama.cpp) at tag `main-b8371-019bda3` | MIT |
-| `tokenizer.json` | [OpenMOSS-Team/MOSS-TTS](https://huggingface.co/OpenMOSS-Team/MOSS-TTS) (Qwen3 backbone tokenizer) | Apache-2.0 |
-| `MOSS-Audio-Tokenizer-ONNX` (encoder + decoder) | [OpenMOSS-Team/MOSS-Audio-Tokenizer-ONNX](https://huggingface.co/OpenMOSS-Team/MOSS-Audio-Tokenizer-ONNX) | Apache-2.0 |
+| **llama-server** | `llama-server/b12110/` | Our CUDA 12.8 build of [llama.cpp](https://github.com/ggml-org/llama.cpp) `b12110` (v0.9.0), merged with its cudart bundle as one flat artifact. The chat + ASR sidecar. |
+| **sd-server** | `sd-server/5dcfa79/` | Our CPU build of [stable-diffusion.cpp](https://github.com/leejet/stable-diffusion.cpp) at commit `5dcfa79` (`master-1104-5dcfa79`). The image-generation sidecar: SD/SDXL/Flux-class diffusion models served over a loopback HTTP API. |
+| **KrisKross** | `kriskross/1.0.0 … 2.1.1/` | OpenBloom's build of the MOSS-TTS inference runtime: the OpenMOSS llama.cpp fork + the MOSS Audio Tokenizer (ONNX). CPU and CUDA 12 variants. |
+| **moss-models** | `moss-models/2.0.0/` | Our GGUF (Q4_K_M) conversion of MOSS-TTS-Realtime 2.0.0, for the KrisKross runtime. |
 
-**Verify before use** — the bundle is hash-pinned by the OpenBloom
-downloader, and you can verify it yourself:
+## Verify before use
 
-kriskross/1.0.0/kriskross-1.0.0-linux-amd64.tar.zst
-sha256: db50ea5521575cbbdcf13a5bfe2ba889d4b44b8b16c482cbec24d5a2d213c9a6
+Every bundle ships a `MANIFEST.json` (the sha256 of each component) and every family's versions
+carry a `SHA256SUMS` file. The consuming installer pins a sha256 and fails closed on any drift —
+you can verify the same way:
 
-Every bundle ships a `MANIFEST.json` listing the sha256 of each
-component, and a `LICENSE` describing the upstream licensing.
+```
+# llama-server (CUDA 12.8, b12110)
+llama-server/b12110/llama-server-b12110-linux-amd64-cuda12.tar.gz
+sha256: 546dab15b51f86a43edd6452a8025e6fba425522102dbb6267cb1a376f7ad8dc
 
-## Attribution
+# sd-server (CPU, stable-diffusion.cpp 5dcfa79)
+sd-server/5dcfa79/sd-server-5dcfa79-linux-amd64-cpu.tar.gz
+sha256: a836547d079fb9c3236d3a6b8344fb93f431a478eb7fe8ba4daf54bfc00c8719
 
-KrisKross is OpenBloom's *build* of components authored by the
-OpenMOSS team and contributors. All credit for the upstream
-llama.cpp fork, the MOSS-TTS models and the MOSS Audio Tokenizer
-belongs to them. OpenBloom Labs produces the compiled bundle and
-publishes it for the OpenBloom installer.
+# KrisKross (latest, 2.1.1, CPU)
+kriskross/2.1.1/kriskross-2.1.1-linux-amd64.tar.zst
+sha256: f2d499917de99eadb856b104f51439544e15fc63a72d273489ecccf1bdc0d4fa
 
-## Integrity policy
+# MOSS-TTS-Realtime (our Q4_K_M conversion)
+moss-models/2.0.0/moss-tts-realtime-q4_k_m.gguf
+sha256: ccd2ab76524affbc3a33a54f56224e24eabb008e73dd2b4142e8bf48cf492ed6
+```
+
+## Upstream attribution & licenses
+
+| Artifact | Built from | License |
+|---|---|---|
+| `llama-server` b12110 | [ggml-org/llama.cpp](https://github.com/ggml-org/llama.cpp) at tag `b12110` | MIT (© The ggml authors) |
+| `sd-server` 5dcfa79 | [leejet/stable-diffusion.cpp](https://github.com/leejet/stable-diffusion.cpp) at commit `5dcfa79` | MIT |
+| `llama-moss-tts` / `llama-moss-tts-realtime` | [OpenBloom/llama.cpp](https://codeberg.org/OpenBloom/llama.cpp) at tag `openbloom-realtime-e1` (the OpenMOSS fork) | MIT |
+| `tokenizer.json` | [OpenMOSS-Team/MOSS-TTS](https://huggingface.co/OpenMOSS-Team/MOSS-TTS) | Apache-2.0 |
+| `MOSS-Audio-Tokenizer-ONNX` | [OpenMOSS-Team/MOSS-Audio-Tokenizer-ONNX](https://huggingface.co/OpenMOSS-Team/MOSS-Audio-Tokenizer-ONNX) | Apache-2.0 |
+| `moss-tts-realtime-q4_k_m.gguf` | [OpenMOSS/MOSS-TTS](https://github.com/OpenMOSS/MOSS-TTS) 2.0.0, quantized by OpenBloom | Apache-2.0 (inherited) |
+
+KrisKross is OpenBloom's *build* of components authored by the OpenMOSS team and contributors;
+all credit for the upstream llama.cpp fork, the MOSS-TTS models and the MOSS Audio Tokenizer
+belongs to them. The sd-server build serves models published by their own authors under their
+own licenses — the models are never hosted here.
 
-- Every artifact in this repository is public and fetchable
-  **unauthenticated**, this is a requirement of the OpenBloom
-  downloader.
-- Every artifact is content-addressed: the consuming installer pins a
-  sha256 and fails closed on any drift.
-- Never gate this repository, and never rewrite a published version
-  in place — publish a new versioned path instead.
+## Integrity policy
 
+- Every artifact in this repository is public and fetchable **unauthenticated** — this is a
+  requirement of the OpenBloom downloader.
+- Every artifact is content-addressed: the consuming installer pins a sha256 and fails closed
+  on any drift.
+- Never gate this repository, and never rewrite a published version in place — publish a new
+  versioned path instead.