From ba2a61f205bb3fcb2474c1c276c9ff4da15cbdee Mon Sep 17 00:00:00 2001 From: localai-org-maint-bot <306269227+localai-org-maint-bot@users.noreply.github.com> Date: Mon, 14 Sep 2026 00:08:16 +0000 Subject: [PATCH] feat(gallery): add Genesis Hermes Final variants Add APEX, Compact, Q8_K_P, and MTP builds with the F16 projector. Pin downloads to one HF revision and verify SHA256 values against LFS metadata and download headers. Document installation and context. Assisted-by: Codex:gpt-6 --- docs/content/features/model-gallery.md | 24 +++ gallery/index.yaml | 238 +++++++++++++++++++++++++ 2 files changed, 262 insertions(+) diff --git a/docs/content/features/model-gallery.md b/docs/content/features/model-gallery.md index f0088f583..6b62bafb8 100644 --- a/docs/content/features/model-gallery.md +++ b/docs/content/features/model-gallery.md @@ -821,3 +821,27 @@ 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. + +### Genesis Hermes Final + +Install the Qwen3.6-35B-A3B Genesis Hermes Final model for text chat, +function calling, and image input: + +```bash +local-ai models install qwen3.6-35b-a3b-genesis-hermes-final +``` + +The gallery offers APEX, APEX Compact, and Q8_K_P builds for llama.cpp. +APEX and APEX Compact also have variants with multi-token prediction (MTP) +enabled. Each build includes the F16 vision projector and uses the embedded +Jinja chat template. LocalAI selects a variant according to available memory +and serving features. To select the plain APEX build explicitly: + +```bash +local-ai models install --variant qwen3.6-35b-a3b-genesis-hermes-final qwen3.6-35b-a3b-genesis-hermes-final +``` + +These entries set a 131,072-token context, following the +[model card's guidance](https://huggingface.co/LuffyTheFox/Qwen3.6-35B-A3B-Uncensored-Genesis-Hermes-Final-GGUF) +for thinking mode. This context requires additional memory beyond the weights. +The uncensored model uses the Apache-2.0 license. diff --git a/gallery/index.yaml b/gallery/index.yaml index 104827cc3..f71405488 100644 --- a/gallery/index.yaml +++ b/gallery/index.yaml @@ -7327,6 +7327,244 @@ - filename: llama-cpp/mmproj/Hermes3.6-35B-A3B-Uncensored-Genesis-V7/mmproj-Hermes3.6-35B-A3B-Uncensored-Genesis-F16.gguf uri: huggingface://LuffyTheFox/Qwen3.6-35B-A3B-Uncensored-Genesis-Hermes-V7-GGUF/mmproj-Hermes3.6-35B-A3B-Uncensored-Genesis-F16.gguf sha256: f7197461d8581cd9be42384d4afe3f851d0844f6d72846d5c97e3dcc931b3cbc +- &qwen3-6-35b-a3b-genesis-hermes-final + name: "qwen3.6-35b-a3b-genesis-hermes-final" + url: "github:mudler/LocalAI/gallery/virtual.yaml@master" + variants: + - model: qwen3.6-35b-a3b-genesis-hermes-final-apex-compact + - model: qwen3.6-35b-a3b-genesis-hermes-final-mtp-apex + - model: qwen3.6-35b-a3b-genesis-hermes-final-mtp-apex-compact + - model: qwen3.6-35b-a3b-genesis-hermes-final-q8-k-p + urls: + - https://huggingface.co/HauhauCS/Qwen3.6-35B-A3B-Uncensored-HauhauCS-Aggressive + - https://huggingface.co/LuffyTheFox/Qwen3.6-35B-A3B-Uncensored-Genesis-Hermes-Final-GGUF + description: | + Qwen3.6-35B-A3B Genesis Hermes Final is a multimodal mixture-of-experts + model with 35B total parameters and about 3B active per token. This + uncensored derivative combines the HauhauCS base with Hermes + function-calling data and the author's Genesis weight processing. + This build uses APEX and includes the F16 vision projector. + license: "apache-2.0" + tags: + - llm + - gguf + - vision + - multimodal + - reasoning + - coding + - uncensored + - moe + last_checked: "2026-09-14" + overrides: + backend: llama-cpp + context_size: 131072 + function: + automatic_tool_parsing_fallback: true + grammar: + disable: true + known_usecases: + - chat + - vision + mmproj: llama-cpp/mmproj/Hermes3.6-35B-A3B-Uncensored-Genesis-Final/mmproj-Hermes3.6-35B-A3B-Uncensored-Genesis-Final-F16.gguf + options: + - use_jinja:true + parameters: + min_p: 0 + model: llama-cpp/models/Hermes3.6-35B-A3B-Uncensored-Genesis-Final-APEX/Hermes3.6-35B-A3B-Uncensored-Genesis-Final-APEX.gguf + repeat_penalty: 1 + temperature: 0.6 + top_k: 20 + top_p: 1 + template: + use_tokenizer_template: true + files: + - filename: llama-cpp/models/Hermes3.6-35B-A3B-Uncensored-Genesis-Final-APEX/Hermes3.6-35B-A3B-Uncensored-Genesis-Final-APEX.gguf + sha256: caefdcec9364e903fc7edac90c39df257f1d6671b5fc389a3cb7dc8ce8204a3b + uri: https://huggingface.co/LuffyTheFox/Qwen3.6-35B-A3B-Uncensored-Genesis-Hermes-Final-GGUF/resolve/d0cf3294f07f2c422f0cf793a75fa48f61d48931/Hermes3.6-35B-A3B-Uncensored-Genesis-Final-APEX.gguf + - filename: llama-cpp/mmproj/Hermes3.6-35B-A3B-Uncensored-Genesis-Final/mmproj-Hermes3.6-35B-A3B-Uncensored-Genesis-Final-F16.gguf + sha256: 5129bb5eb19e4346c0f2071f1ce8e1b0a076e0ab08d57a77d6033ce01235252c + uri: https://huggingface.co/LuffyTheFox/Qwen3.6-35B-A3B-Uncensored-Genesis-Hermes-Final-GGUF/resolve/d0cf3294f07f2c422f0cf793a75fa48f61d48931/mmproj-Hermes3.6-35B-A3B-Uncensored-Genesis-Final-F16.gguf +- !!merge <<: *qwen3-6-35b-a3b-genesis-hermes-final + name: "qwen3.6-35b-a3b-genesis-hermes-final-apex-compact" + variants: [] + description: | + Qwen3.6-35B-A3B Genesis Hermes Final is a multimodal mixture-of-experts + model with 35B total parameters and about 3B active per token. This + uncensored derivative combines the HauhauCS base with Hermes + function-calling data and the author's Genesis weight processing. + This build uses APEX Compact and includes the F16 vision projector. + overrides: + backend: llama-cpp + context_size: 131072 + function: + automatic_tool_parsing_fallback: true + grammar: + disable: true + known_usecases: + - chat + - vision + mmproj: llama-cpp/mmproj/Hermes3.6-35B-A3B-Uncensored-Genesis-Final/mmproj-Hermes3.6-35B-A3B-Uncensored-Genesis-Final-F16.gguf + options: + - use_jinja:true + parameters: + min_p: 0 + model: llama-cpp/models/Hermes3.6-35B-A3B-Uncensored-Genesis-Final-APEX-Compact/Hermes3.6-35B-A3B-Uncensored-Genesis-Final-APEX-Compact.gguf + repeat_penalty: 1 + temperature: 0.6 + top_k: 20 + top_p: 1 + template: + use_tokenizer_template: true + files: + - filename: llama-cpp/models/Hermes3.6-35B-A3B-Uncensored-Genesis-Final-APEX-Compact/Hermes3.6-35B-A3B-Uncensored-Genesis-Final-APEX-Compact.gguf + sha256: c55f3a0acc94d28a192cf1262a9cc5dbdbe15ce7bad75894ce2b0d7e8b66ddb8 + uri: https://huggingface.co/LuffyTheFox/Qwen3.6-35B-A3B-Uncensored-Genesis-Hermes-Final-GGUF/resolve/d0cf3294f07f2c422f0cf793a75fa48f61d48931/Hermes3.6-35B-A3B-Uncensored-Genesis-Final-APEX-Compact.gguf + - filename: llama-cpp/mmproj/Hermes3.6-35B-A3B-Uncensored-Genesis-Final/mmproj-Hermes3.6-35B-A3B-Uncensored-Genesis-Final-F16.gguf + sha256: 5129bb5eb19e4346c0f2071f1ce8e1b0a076e0ab08d57a77d6033ce01235252c + uri: https://huggingface.co/LuffyTheFox/Qwen3.6-35B-A3B-Uncensored-Genesis-Hermes-Final-GGUF/resolve/d0cf3294f07f2c422f0cf793a75fa48f61d48931/mmproj-Hermes3.6-35B-A3B-Uncensored-Genesis-Final-F16.gguf +- !!merge <<: *qwen3-6-35b-a3b-genesis-hermes-final + name: "qwen3.6-35b-a3b-genesis-hermes-final-mtp-apex" + variants: [] + description: | + Qwen3.6-35B-A3B Genesis Hermes Final is a multimodal mixture-of-experts + model with 35B total parameters and about 3B active per token. This + uncensored derivative combines the HauhauCS base with Hermes + function-calling data and the author's Genesis weight processing. + This build uses APEX and includes the F16 vision projector. + Native multi-token prediction is enabled for speculative decoding. + tags: + - llm + - gguf + - vision + - multimodal + - reasoning + - coding + - uncensored + - moe + - mtp + overrides: + backend: llama-cpp + context_size: 131072 + function: + automatic_tool_parsing_fallback: true + grammar: + disable: true + known_usecases: + - chat + - vision + mmproj: llama-cpp/mmproj/Hermes3.6-35B-A3B-Uncensored-Genesis-Final/mmproj-Hermes3.6-35B-A3B-Uncensored-Genesis-Final-F16.gguf + options: + - use_jinja:true + - spec_type:draft-mtp + - spec_n_max:6 + - spec_p_min:0.75 + parameters: + min_p: 0 + model: llama-cpp/models/Hermes3.6-35B-A3B-Uncensored-Genesis-Final-MTP-APEX/Hermes3.6-35B-A3B-Uncensored-Genesis-Final-MTP-APEX.gguf + repeat_penalty: 1 + temperature: 0.6 + top_k: 20 + top_p: 1 + template: + use_tokenizer_template: true + files: + - filename: llama-cpp/models/Hermes3.6-35B-A3B-Uncensored-Genesis-Final-MTP-APEX/Hermes3.6-35B-A3B-Uncensored-Genesis-Final-MTP-APEX.gguf + sha256: 01e16a28774e4f075a2bfd92ff889cb7b004e37511e68399caf39a3f4b195722 + uri: https://huggingface.co/LuffyTheFox/Qwen3.6-35B-A3B-Uncensored-Genesis-Hermes-Final-GGUF/resolve/d0cf3294f07f2c422f0cf793a75fa48f61d48931/Hermes3.6-35B-A3B-Uncensored-Genesis-Final-MTP-APEX.gguf + - filename: llama-cpp/mmproj/Hermes3.6-35B-A3B-Uncensored-Genesis-Final/mmproj-Hermes3.6-35B-A3B-Uncensored-Genesis-Final-F16.gguf + sha256: 5129bb5eb19e4346c0f2071f1ce8e1b0a076e0ab08d57a77d6033ce01235252c + uri: https://huggingface.co/LuffyTheFox/Qwen3.6-35B-A3B-Uncensored-Genesis-Hermes-Final-GGUF/resolve/d0cf3294f07f2c422f0cf793a75fa48f61d48931/mmproj-Hermes3.6-35B-A3B-Uncensored-Genesis-Final-F16.gguf +- !!merge <<: *qwen3-6-35b-a3b-genesis-hermes-final + name: "qwen3.6-35b-a3b-genesis-hermes-final-mtp-apex-compact" + variants: [] + description: | + Qwen3.6-35B-A3B Genesis Hermes Final is a multimodal mixture-of-experts + model with 35B total parameters and about 3B active per token. This + uncensored derivative combines the HauhauCS base with Hermes + function-calling data and the author's Genesis weight processing. + This build uses APEX Compact and includes the F16 vision projector. + Native multi-token prediction is enabled for speculative decoding. + tags: + - llm + - gguf + - vision + - multimodal + - reasoning + - coding + - uncensored + - moe + - mtp + overrides: + backend: llama-cpp + context_size: 131072 + function: + automatic_tool_parsing_fallback: true + grammar: + disable: true + known_usecases: + - chat + - vision + mmproj: llama-cpp/mmproj/Hermes3.6-35B-A3B-Uncensored-Genesis-Final/mmproj-Hermes3.6-35B-A3B-Uncensored-Genesis-Final-F16.gguf + options: + - use_jinja:true + - spec_type:draft-mtp + - spec_n_max:6 + - spec_p_min:0.75 + parameters: + min_p: 0 + model: llama-cpp/models/Hermes3.6-35B-A3B-Uncensored-Genesis-Final-MTP-APEX-Compact/Hermes3.6-35B-A3B-Uncensored-Genesis-Final-MTP-APEX-Compact.gguf + repeat_penalty: 1 + temperature: 0.6 + top_k: 20 + top_p: 1 + template: + use_tokenizer_template: true + files: + - filename: llama-cpp/models/Hermes3.6-35B-A3B-Uncensored-Genesis-Final-MTP-APEX-Compact/Hermes3.6-35B-A3B-Uncensored-Genesis-Final-MTP-APEX-Compact.gguf + sha256: fe921f9f7d3f48f9fbda1f28bb3e96824db051958cbc6e44dd0e5bebc690829a + uri: https://huggingface.co/LuffyTheFox/Qwen3.6-35B-A3B-Uncensored-Genesis-Hermes-Final-GGUF/resolve/d0cf3294f07f2c422f0cf793a75fa48f61d48931/Hermes3.6-35B-A3B-Uncensored-Genesis-Final-MTP-APEX-Compact.gguf + - filename: llama-cpp/mmproj/Hermes3.6-35B-A3B-Uncensored-Genesis-Final/mmproj-Hermes3.6-35B-A3B-Uncensored-Genesis-Final-F16.gguf + sha256: 5129bb5eb19e4346c0f2071f1ce8e1b0a076e0ab08d57a77d6033ce01235252c + uri: https://huggingface.co/LuffyTheFox/Qwen3.6-35B-A3B-Uncensored-Genesis-Hermes-Final-GGUF/resolve/d0cf3294f07f2c422f0cf793a75fa48f61d48931/mmproj-Hermes3.6-35B-A3B-Uncensored-Genesis-Final-F16.gguf +- !!merge <<: *qwen3-6-35b-a3b-genesis-hermes-final + name: "qwen3.6-35b-a3b-genesis-hermes-final-q8-k-p" + variants: [] + description: | + Qwen3.6-35B-A3B Genesis Hermes Final is a multimodal mixture-of-experts + model with 35B total parameters and about 3B active per token. This + uncensored derivative combines the HauhauCS base with Hermes + function-calling data and the author's Genesis weight processing. + This build uses Q8_K_P and includes the F16 vision projector. + overrides: + backend: llama-cpp + context_size: 131072 + function: + automatic_tool_parsing_fallback: true + grammar: + disable: true + known_usecases: + - chat + - vision + mmproj: llama-cpp/mmproj/Hermes3.6-35B-A3B-Uncensored-Genesis-Final/mmproj-Hermes3.6-35B-A3B-Uncensored-Genesis-Final-F16.gguf + options: + - use_jinja:true + parameters: + min_p: 0 + model: llama-cpp/models/Hermes3.6-35B-A3B-Uncensored-Genesis-Final-Q8_K_P/Hermes3.6-35B-A3B-Uncensored-Genesis-Final-Q8_K_P.gguf + repeat_penalty: 1 + temperature: 0.6 + top_k: 20 + top_p: 1 + template: + use_tokenizer_template: true + files: + - filename: llama-cpp/models/Hermes3.6-35B-A3B-Uncensored-Genesis-Final-Q8_K_P/Hermes3.6-35B-A3B-Uncensored-Genesis-Final-Q8_K_P.gguf + sha256: e1ac7246dd3ebfb036732874dc57acbacc8b0d7d02df4417e2ac0d1f376e8d1d + uri: https://huggingface.co/LuffyTheFox/Qwen3.6-35B-A3B-Uncensored-Genesis-Hermes-Final-GGUF/resolve/d0cf3294f07f2c422f0cf793a75fa48f61d48931/Hermes3.6-35B-A3B-Uncensored-Genesis-Final-Q8_K_P.gguf + - filename: llama-cpp/mmproj/Hermes3.6-35B-A3B-Uncensored-Genesis-Final/mmproj-Hermes3.6-35B-A3B-Uncensored-Genesis-Final-F16.gguf + sha256: 5129bb5eb19e4346c0f2071f1ce8e1b0a076e0ab08d57a77d6033ce01235252c + uri: https://huggingface.co/LuffyTheFox/Qwen3.6-35B-A3B-Uncensored-Genesis-Hermes-Final-GGUF/resolve/d0cf3294f07f2c422f0cf793a75fa48f61d48931/mmproj-Hermes3.6-35B-A3B-Uncensored-Genesis-Final-F16.gguf + - &kat-coder-v2-5-dev name: "kat-coder-v2.5-dev" variants: