[UPDATE][llamacppllmbasev3][1.2.33]

llamacppllmbasev3/Chart.yaml

@@ -1,6 +1,6 @@
 apiVersion: v2
-appVersion: b10362
+appVersion: b10454
 description: Generic llama.cpp + llm-init base; the GGUF model is supplied at install time via env
 name: llamacppllmbasev3
 type: application
-version: 1.2.30
+version: 1.2.33

llamacppllmbasev3/OlaresManifest.yaml

@@ -7,9 +7,11 @@ metadata:
   description: "Generic llama.cpp engine base. Pick any GGUF model at install via env."
   appid: llamacppllmbasev3
   title: llama.cpp Engine Base
-  version: '1.2.30'
+  version: '1.2.33'
   categories:
     - AI
+  tags:
+    - engine
 sharedEntrances:
   - name: llamacppllmbasev3
     host: sharedentrances-api
@@ -33,17 +35,25 @@ workloadReplicas:
   llminit: 1
 spec:
   onlyAdmin: true
-  versionName: 'b10362'
+  versionName: 'b10454'
   upgradeDescription: |
-    Chart 1.2.30 — clarify upgrade notes for engine `server-cuda12-b10362` (unchanged from 1.2.29).
+    Chart 1.2.33 — bump engine to `server-cuda12-b10454` and llm-init to `v1.3.8`.
 
-    **Muse Glimmer 30B (Meta)** — requires **b10353+**; **b10362 includes** the `muse-glimmer` architecture ([llama.cpp #26841](https://github.com/ggml-org/llama.cpp/pull/26841)). Older builds (e.g. b10331) fail with `unknown model architecture: 'muse-glimmer'`.
+    **Images**
+    - llama.cpp: `docker.io/beclab/ggml-org-llama.cpp:server-cuda12-b10362` → `server-cuda12-b10454` (latest published beclab CUDA 12 server tag; amd64 + arm64). Upstream GitHub is already at b10456; no beclab b10455/b10456 image yet.
+    - llm-init: `docker.io/beclab/llm-init:v1.3.5` → `v1.3.8`.
+
+    **Muse Glimmer 30B (Meta)** — requires **b10353+**; **b10454 includes** the `muse-glimmer` architecture ([llama.cpp #26841](https://github.com/ggml-org/llama.cpp/pull/26841)). Older builds (e.g. b10331) fail with `unknown model architecture: 'muse-glimmer'`.
 
     Clone this base and point env at [meta-models/Muse-Glimmer-30B-GGUF](https://huggingface.co/meta-models/Muse-Glimmer-30B-GGUF) or community quants (e.g. unsloth). Typical layout:
     - Main GGUF + `mmproj` in `MODEL_SOURCE` (comma-separated) for vision
     - DFlash drafter (`dflash-kquant.gguf`) + `--spec-type draft-dflash` or `draft-dspark` in `ENGINE_ARGS` for speculative decode
 
-    **Also in b10331 → b10362:** DSpark / Eagle3-v3 / MTP spec-decode, DeepSeek V4 & multimodal parsers, CUDA Q2_0 / Mamba-2 SSD prefill, server router & metrics improvements.
+    **Also in b10362 → b10454:**
+    - Chat templates receive OpenAI `reasoning_effort` ([llama.cpp #26941](https://github.com/ggml-org/llama.cpp/pull/26941)); Qwen3.8 uses `xhigh` / `medium` / `low` via `--chat-template-kwargs` or the request body
+    - MTP draft type auto-detect; DFlash / DSpark backend sampling
+    - MiniMax Text01 / M1 conversion support
+    - Server `/metrics` and `/slots` stay available during decode; mtmd and jinja fixes
 
     See https://github.com/ggml-org/llama.cpp/releases for full upstream release notes.
   fullDescription: |

llamacppllmbasev3/i18n/en-US/OlaresManifest.yaml

@@ -3,6 +3,26 @@ metadata:
   title: llama.cpp Engine Base
 
 spec:
+  upgradeDescription: |
+    Chart 1.2.33 — bump engine to `server-cuda12-b10454` and llm-init to `v1.3.8`.
+
+    **Images**
+    - llama.cpp: `docker.io/beclab/ggml-org-llama.cpp:server-cuda12-b10362` → `server-cuda12-b10454` (latest published beclab CUDA 12 server tag; amd64 + arm64). Upstream GitHub is already at b10456; no beclab b10455/b10456 image yet.
+    - llm-init: `docker.io/beclab/llm-init:v1.3.5` → `v1.3.8`.
+
+    **Muse Glimmer 30B (Meta)** — requires **b10353+**; **b10454 includes** the `muse-glimmer` architecture ([llama.cpp #26841](https://github.com/ggml-org/llama.cpp/pull/26841)). Older builds (e.g. b10331) fail with `unknown model architecture: 'muse-glimmer'`.
+
+    Clone this base and point env at [meta-models/Muse-Glimmer-30B-GGUF](https://huggingface.co/meta-models/Muse-Glimmer-30B-GGUF) or community quants (e.g. unsloth). Typical layout:
+    - Main GGUF + `mmproj` in `MODEL_SOURCE` (comma-separated) for vision
+    - DFlash drafter (`dflash-kquant.gguf`) + `--spec-type draft-dflash` or `draft-dspark` in `ENGINE_ARGS` for speculative decode
+
+    **Also in b10362 → b10454:**
+    - Chat templates receive OpenAI `reasoning_effort` ([llama.cpp #26941](https://github.com/ggml-org/llama.cpp/pull/26941)); Qwen3.8 uses `xhigh` / `medium` / `low` via `--chat-template-kwargs` or the request body
+    - MTP draft type auto-detect; DFlash / DSpark backend sampling
+    - MiniMax Text01 / M1 conversion support
+    - Server `/metrics` and `/slots` stay available during decode; mtmd and jinja fixes
+
+    See https://github.com/ggml-org/llama.cpp/releases for full upstream release notes.
   fullDescription: |
     **IMPORTANT NOTE**  
     This app is a template and cannot be used on its own. To use it, set the environment variables below to connect a specific model.

llamacppllmbasev3/i18n/zh-CN/OlaresManifest.yaml

@@ -3,6 +3,26 @@ metadata:
   title: llama.cpp 引擎基座
 
 spec:
+  upgradeDescription: |
+    Chart 1.2.33 — 引擎升级到 `server-cuda12-b10454`，llm-init 升级到 `v1.3.8`。
+
+    **镜像**
+    - llama.cpp：`docker.io/beclab/ggml-org-llama.cpp:server-cuda12-b10362` → `server-cuda12-b10454`（当前已发布的最新 beclab CUDA 12 server 标签；amd64 + arm64）。上游 GitHub 已到 b10456；尚无 beclab 的 b10455/b10456 镜像。
+    - llm-init：`docker.io/beclab/llm-init:v1.3.5` → `v1.3.8`。
+
+    **Muse Glimmer 30B（Meta）** — 需要 **b10353+**；**b10454 已包含** `muse-glimmer` 架构（[llama.cpp #26841](https://github.com/ggml-org/llama.cpp/pull/26841)）。更旧的构建（如 b10331）会报 `unknown model architecture: 'muse-glimmer'`。
+
+    克隆此基座后，将环境变量指向 [meta-models/Muse-Glimmer-30B-GGUF](https://huggingface.co/meta-models/Muse-Glimmer-30B-GGUF) 或社区量化（如 unsloth）。常见布局：
+    - 主 GGUF + `mmproj` 写入 `MODEL_SOURCE`（逗号分隔）以支持视觉
+    - DFlash 草稿模型（`dflash-kquant.gguf`）并在 `ENGINE_ARGS` 中加 `--spec-type draft-dflash` 或 `draft-dspark` 以开启投机解码
+
+    **b10362 → b10454 其它变化：**
+    - 聊天模板可接收 OpenAI `reasoning_effort`（[llama.cpp #26941](https://github.com/ggml-org/llama.cpp/pull/26941)）；Qwen3.8 可用 `--chat-template-kwargs` 或请求体设置 `xhigh` / `medium` / `low`
+    - MTP 草稿类型自动检测；DFlash / DSpark 后端采样
+    - 支持 MiniMax Text01 / M1 转换
+    - 解码过程中 `/metrics` 与 `/slots` 仍可用；mtmd 与 jinja 修复
+
+    完整上游发布说明见 https://github.com/ggml-org/llama.cpp/releases
   fullDescription: |
     **重要提示**  
     此应用是一个模板，不能单独使用。请按下方环境变量设置模型后，方可使用本应用提供模型服务。

llamacppllmbasev3/templates/llamacpp.yaml

@@ -73,7 +73,7 @@ spec:
             defaultMode: 0555
       containers:
         - name: llamacpp
-          image: docker.io/beclab/ggml-org-llama.cpp:server-cuda12-b10362
+          image: docker.io/beclab/ggml-org-llama.cpp:server-cuda12-b10454
           imagePullPolicy: IfNotPresent
           command: ["/bin/sh", "/llm-init/wrappers/llamacpp.sh"]
           env:

llamacppllmbasev3/templates/llm-init.yaml

@@ -109,7 +109,7 @@ spec:
           # v1.2.4 defaults to the stable LFS download path (hf_xet disabled
           # unless HF_ENABLE_XET=true); LFS streams each file to disk with a
           # bounded RAM footprint, avoiding the hf_xet OOM (huggingface_hub#3300).
-          image: docker.io/beclab/llm-init:v1.3.5
+          image: docker.io/beclab/llm-init:v1.3.8
           imagePullPolicy: IfNotPresent
           env:
             - name: ENGINE_KIND