From a5accc7488bacebbce52f8d9ac26e4b9e6a00d88 Mon Sep 17 00:00:00 2001 From: localai-org-maint-bot <306269227+localai-org-maint-bot@users.noreply.github.com> Date: Thu, 8 Oct 2026 20:03:54 +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 --- docs/content/features/model-gallery.md | 24 ++++++++ gallery/index.yaml | 76 ++++++++++++++++++++++++++ 2 files changed, 100 insertions(+) diff --git a/docs/content/features/model-gallery.md b/docs/content/features/model-gallery.md index b665be21c..cea590268 100644 --- a/docs/content/features/model-gallery.md +++ b/docs/content/features/model-gallery.md @@ -902,3 +902,27 @@ is still running is reported as queued until the installer picks it up: A job ID is queryable from the moment `/models/apply` returns it, so a `404`/`500` from this endpoint means the ID is genuinely unknown rather than merely waiting its turn. + +### Mellum2.1 Thinking + +[Mellum2.1 Thinking](https://huggingface.co/JetBrains/Mellum2.1-12B-A2.5B-Thinking) +is a JetBrains reasoning model with 12 billion total parameters and 2.5 billion +active parameters per token. The gallery provides Q4_K_M and Q8_0 GGUF builds +for the llama-cpp backend. + +Install the default entry with automatic variant selection: + +```bash +local-ai models install mellum2.1-12b-a2.5b-thinking +``` + +To select Q8_0 explicitly: + +```bash +local-ai models install mellum2.1-12b-a2.5b-thinking --variant mellum2.1-12b-a2.5b-thinking-q8 +``` + +Both entries use the embedded chat template, temperature 0.6, top-p 0.95, +and top-k 20. The default context is 32,768 tokens; the model supports up to +131,072 tokens. Increasing the context requires more memory. The model emits +reasoning inside `...` blocks before its final answer. diff --git a/gallery/index.yaml b/gallery/index.yaml index 17a7cdfa1..80d24680f 100644 --- a/gallery/index.yaml +++ b/gallery/index.yaml @@ -8766,6 +8766,82 @@ - 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: