From e8e28a98f915a8c206f298f0d6a692be8341e04e Mon Sep 17 00:00:00 2001
From: localai-org-maint-bot
 <306269227+localai-org-maint-bot@users.noreply.github.com>
Date: Tue, 13 Oct 2026 07:27:00 +0000
Subject: [PATCH] chore(gallery): add Mellum2.1 Thinking variants

Add JetBrains Q4_K_M and Q8_0 GGUF builds for llama.cpp, with
variant selection and installation documentation. Pin both downloads
and verify their SHA256 values against LFS metadata and HTTP headers.

Assisted-by: Codex:gpt-6

(cherry picked from commit a5accc7488bacebbce52f8d9ac26e4b9e6a00d88, mudler/LocalAI#12573)
Signed-off-by: Ettore Di Giacinto <mudler@localai.io>
---
 gallery/index.yaml | 76 ++++++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 76 insertions(+)

diff --git a/gallery/index.yaml b/gallery/index.yaml
index 56e4d2a66..eaf6f08da 100644
--- a/gallery/index.yaml
+++ b/gallery/index.yaml
@@ -9334,6 +9334,82 @@
     - filename: llama-cpp/models/FINAL-Bench/POCKET-26B-GGUF/POCKET-26B-Q2_K.gguf
       sha256: 490687be6493769dc8322df396cd31d412a15877eaccea4d2f16ca1f93383334
       uri: huggingface://FINAL-Bench/POCKET-26B-GGUF/POCKET-26B-Q2_K.gguf
+- name: mellum2.1-12b-a2.5b-thinking
+  variants:
+    - model: mellum2.1-12b-a2.5b-thinking-q8
+  url: github:mudler/LocalAI/gallery/virtual.yaml@master
+  urls:
+    - https://huggingface.co/JetBrains/Mellum2.1-12B-A2.5B-Thinking
+    - https://huggingface.co/JetBrains/Mellum2.1-12B-A2.5B-Thinking-GGUF
+  description: |
+    Mellum2.1 Thinking is a JetBrains mixture-of-experts reasoning model with
+    12 billion total parameters and 2.5 billion active parameters per token.
+    It emits reasoning in think blocks and supports a 131,072-token context.
+    This entry uses Q4_K_M GGUF weights with a 32,768-token default context.
+  license: apache-2.0
+  tags:
+    - llm
+    - gguf
+    - cpu
+    - gpu
+    - moe
+    - reasoning
+  last_checked: "2026-10-08"
+  overrides:
+    backend: llama-cpp
+    context_size: 32768
+    known_usecases:
+      - chat
+    options:
+      - use_jinja:true
+    parameters:
+      model: llama-cpp/models/JetBrains/Mellum2.1-12B-A2.5B-Thinking/Mellum2.1-12B-A2.5B-Thinking-Q4_K_M.gguf
+      temperature: 0.6
+      top_p: 0.95
+      top_k: 20
+    template:
+      use_tokenizer_template: true
+  files:
+    - filename: llama-cpp/models/JetBrains/Mellum2.1-12B-A2.5B-Thinking/Mellum2.1-12B-A2.5B-Thinking-Q4_K_M.gguf
+      sha256: ecc4d5b8107fc219e23c494b2d107552137637b6c6002ee713e0ba8af7fa2755
+      uri: https://huggingface.co/JetBrains/Mellum2.1-12B-A2.5B-Thinking-GGUF/resolve/20b439426f1e997ba575eaccaf4856ac0f76f5d5/Mellum2.1-12B-A2.5B-Thinking-Q4_K_M.gguf
+- name: mellum2.1-12b-a2.5b-thinking-q8
+  url: github:mudler/LocalAI/gallery/virtual.yaml@master
+  urls:
+    - https://huggingface.co/JetBrains/Mellum2.1-12B-A2.5B-Thinking
+    - https://huggingface.co/JetBrains/Mellum2.1-12B-A2.5B-Thinking-GGUF
+  description: |
+    Mellum2.1 Thinking is a JetBrains mixture-of-experts reasoning model with
+    12 billion total parameters and 2.5 billion active parameters per token.
+    It emits reasoning in think blocks and supports a 131,072-token context.
+    This entry uses Q8_0 GGUF weights with a 32,768-token default context.
+  license: apache-2.0
+  tags:
+    - llm
+    - gguf
+    - cpu
+    - gpu
+    - moe
+    - reasoning
+  last_checked: "2026-10-08"
+  overrides:
+    backend: llama-cpp
+    context_size: 32768
+    known_usecases:
+      - chat
+    options:
+      - use_jinja:true
+    parameters:
+      model: llama-cpp/models/JetBrains/Mellum2.1-12B-A2.5B-Thinking/Mellum2.1-12B-A2.5B-Thinking-Q8_0.gguf
+      temperature: 0.6
+      top_p: 0.95
+      top_k: 20
+    template:
+      use_tokenizer_template: true
+  files:
+    - filename: llama-cpp/models/JetBrains/Mellum2.1-12B-A2.5B-Thinking/Mellum2.1-12B-A2.5B-Thinking-Q8_0.gguf
+      sha256: 69e0ac02a16586c2c9e991f1e3b8a1eb403882e7ab9e69ea01eb23dbbbc6841f
+      uri: https://huggingface.co/JetBrains/Mellum2.1-12B-A2.5B-Thinking-GGUF/resolve/20b439426f1e997ba575eaccaf4856ac0f76f5d5/Mellum2.1-12B-A2.5B-Thinking-Q8_0.gguf
 - &mellum2-12b-a2-5b-instruct
   name: mellum2-12b-a2.5b-instruct
   variants: