From 5fbaaaea2a671ded851cebf4fc75415420459406 Mon Sep 17 00:00:00 2001 From: localai-org-maint-bot <306269227+localai-org-maint-bot@users.noreply.github.com> Date: Sun, 25 Oct 2026 09:53:03 +0000 Subject: [PATCH] feat(gallery): add Flash Next Turbo Brilliance Add the IQ3_S GGUF build with both shards and its vision projector. Pin all downloads and document the installation command. Assisted-by: Codex:GPT-6 (cherry picked from commit 15ed1cb4e1588e78cfd8e9c4517ab73abfe6e01e, mudler/LocalAI#12860) Assisted-by: Claude:claude-sonnet-5-5 Signed-off-by: Ettore Di Giacinto --- gallery/index.yaml | 65 ++++++++++++++++++++++++++++++++++ gallery/models/index.yaml | 73 +++++++++++++++++++++++++++++++++++++++ 2 files changed, 138 insertions(+) diff --git a/gallery/index.yaml b/gallery/index.yaml index b6c2b68ce..97f2d048a 100644 --- a/gallery/index.yaml +++ b/gallery/index.yaml @@ -4838,6 +4838,71 @@ template: use_tokenizer_template: true variants: [] +- url: github:mudler/LocalAI/gallery/virtual.yaml@master + name: qwen3.8-flash-next-turbo-brilliance-iq3-s + description: | + DavidAU's Turbo Brilliance build of Qwen3.8 Flash Next uses GSQ-RCO + IQ3_S mixed quantization and the publisher's embedded chat template. + Includes both GGUF shards and the BF16 vision projector for text chat + and image input, with a 32K default context. See the model card for + the publisher's reasoning and instruction modes. + license: other + urls: + - https://huggingface.co/Qwen/Qwen3.8-Flash-Next + - https://huggingface.co/ISTA-DASLab/Qwen3.8-Flash-Next-GSQ-RCO-GGUF + - https://huggingface.co/DavidAU/Qwen3.8-Flash-Next-Turbo-Brilliance-50x-Reasoning-Instruct-GSQ-RCO-GGUF + icon: https://qianwen-res.oss-cn-beijing.aliyuncs.com/logo_qwen.jpg + tags: + - llm + - gguf + - cpu + - gpu + - qwen + - moe + - reasoning + - thinking + - coding + - agent + - tools + - vision + - multimodal + - long-context + files: + - filename: llama-cpp/models/qwen3.8-flash-next-turbo-brilliance-iq3-s/Qwen3.8-Flash-Next-GSQ-RCO-IQ3_S-00001-of-00002.gguf + sha256: 8a14ab07ca7b54619e6b250cd1d83f97f79ab7eef3caf8ee9e6543bc9ac6a193 + uri: https://huggingface.co/DavidAU/Qwen3.8-Flash-Next-Turbo-Brilliance-50x-Reasoning-Instruct-GSQ-RCO-GGUF/resolve/8d09a8a84cc067203d2903db831c8aa8713a5f3f/Qwen3.8-Flash-Next-GSQ-RCO-IQ3_S-00001-of-00002.gguf + - filename: llama-cpp/models/qwen3.8-flash-next-turbo-brilliance-iq3-s/Qwen3.8-Flash-Next-GSQ-RCO-IQ3_S-00002-of-00002.gguf + sha256: 316b46f3a2dbd68c900f43136ab9449f9dcc3725dfd8c794847c204bc161e113 + uri: https://huggingface.co/DavidAU/Qwen3.8-Flash-Next-Turbo-Brilliance-50x-Reasoning-Instruct-GSQ-RCO-GGUF/resolve/8d09a8a84cc067203d2903db831c8aa8713a5f3f/Qwen3.8-Flash-Next-GSQ-RCO-IQ3_S-00002-of-00002.gguf + - filename: llama-cpp/mmproj/qwen3.8-flash-next-turbo-brilliance-iq3-s/mmproj-Qwen3.8-Flash-Next-BF16.gguf + sha256: b1a82259702816a5330d7bd7607cd9676b11780e79ff7348c21103ff3ce49bd0 + uri: https://huggingface.co/DavidAU/Qwen3.8-Flash-Next-Turbo-Brilliance-50x-Reasoning-Instruct-GSQ-RCO-GGUF/resolve/8d09a8a84cc067203d2903db831c8aa8713a5f3f/mmproj-Qwen3.8-Flash-Next-BF16.gguf + overrides: + backend: llama-cpp + context_size: 32768 + function: + automatic_tool_parsing_fallback: true + grammar: + disable: true + known_usecases: + - chat + - vision + mlock: false + mmap: true + mmproj: llama-cpp/mmproj/qwen3.8-flash-next-turbo-brilliance-iq3-s/mmproj-Qwen3.8-Flash-Next-BF16.gguf + options: + - use_jinja:true + - fit:false + parameters: + min_p: 0 + model: llama-cpp/models/qwen3.8-flash-next-turbo-brilliance-iq3-s/Qwen3.8-Flash-Next-GSQ-RCO-IQ3_S-00001-of-00002.gguf + presence_penalty: 0 + repeat_penalty: 1 + temperature: 1 + top_k: 20 + top_p: 0.95 + template: + use_tokenizer_template: true - url: github:mudler/LocalAI/gallery/virtual.yaml@master name: qwen3.8-flash-next-gsq-rco-q2-0 description: | diff --git a/gallery/models/index.yaml b/gallery/models/index.yaml index ecdf697d6..76b9ccc47 100644 --- a/gallery/models/index.yaml +++ b/gallery/models/index.yaml @@ -5671,6 +5671,79 @@ template: use_tokenizer_template: true variants: [] +- kind: model + id: qwen3.8-flash-next-turbo-brilliance-iq3-s + name: qwen3.8-flash-next-turbo-brilliance-iq3-s + version: 0.0.0-gallery.e69b0c67b3e4 + description: | + DavidAU's Turbo Brilliance build of Qwen3.8 Flash Next uses GSQ-RCO + IQ3_S mixed quantization and the publisher's embedded chat template. + Includes both GGUF shards and the BF16 vision projector for text chat + and image input, with a 32K default context. See the model card for + the publisher's reasoning and instruction modes. + license: other + urls: + - https://huggingface.co/Qwen/Qwen3.8-Flash-Next + - https://huggingface.co/ISTA-DASLab/Qwen3.8-Flash-Next-GSQ-RCO-GGUF + - https://huggingface.co/DavidAU/Qwen3.8-Flash-Next-Turbo-Brilliance-50x-Reasoning-Instruct-GSQ-RCO-GGUF + tags: + - llm + - gguf + - cpu + - gpu + - qwen + - moe + - reasoning + - thinking + - coding + - agent + - tools + - vision + - multimodal + - long-context + icon: https://qianwen-res.oss-cn-beijing.aliyuncs.com/logo_qwen.jpg + channel: gallery + policy: + requireSignature: false + schemaVersion: 1 + model: + url: github:mudler/LocalAI/gallery/virtual.yaml@master + files: + - filename: llama-cpp/models/qwen3.8-flash-next-turbo-brilliance-iq3-s/Qwen3.8-Flash-Next-GSQ-RCO-IQ3_S-00001-of-00002.gguf + sha256: 8a14ab07ca7b54619e6b250cd1d83f97f79ab7eef3caf8ee9e6543bc9ac6a193 + uri: https://huggingface.co/DavidAU/Qwen3.8-Flash-Next-Turbo-Brilliance-50x-Reasoning-Instruct-GSQ-RCO-GGUF/resolve/8d09a8a84cc067203d2903db831c8aa8713a5f3f/Qwen3.8-Flash-Next-GSQ-RCO-IQ3_S-00001-of-00002.gguf + - filename: llama-cpp/models/qwen3.8-flash-next-turbo-brilliance-iq3-s/Qwen3.8-Flash-Next-GSQ-RCO-IQ3_S-00002-of-00002.gguf + sha256: 316b46f3a2dbd68c900f43136ab9449f9dcc3725dfd8c794847c204bc161e113 + uri: https://huggingface.co/DavidAU/Qwen3.8-Flash-Next-Turbo-Brilliance-50x-Reasoning-Instruct-GSQ-RCO-GGUF/resolve/8d09a8a84cc067203d2903db831c8aa8713a5f3f/Qwen3.8-Flash-Next-GSQ-RCO-IQ3_S-00002-of-00002.gguf + - filename: llama-cpp/mmproj/qwen3.8-flash-next-turbo-brilliance-iq3-s/mmproj-Qwen3.8-Flash-Next-BF16.gguf + sha256: b1a82259702816a5330d7bd7607cd9676b11780e79ff7348c21103ff3ce49bd0 + uri: https://huggingface.co/DavidAU/Qwen3.8-Flash-Next-Turbo-Brilliance-50x-Reasoning-Instruct-GSQ-RCO-GGUF/resolve/8d09a8a84cc067203d2903db831c8aa8713a5f3f/mmproj-Qwen3.8-Flash-Next-BF16.gguf + overrides: + backend: llama-cpp + context_size: 32768 + function: + automatic_tool_parsing_fallback: true + grammar: + disable: true + known_usecases: + - chat + - vision + mlock: false + mmap: true + mmproj: llama-cpp/mmproj/qwen3.8-flash-next-turbo-brilliance-iq3-s/mmproj-Qwen3.8-Flash-Next-BF16.gguf + options: + - use_jinja:true + - fit:false + parameters: + min_p: 0 + model: llama-cpp/models/qwen3.8-flash-next-turbo-brilliance-iq3-s/Qwen3.8-Flash-Next-GSQ-RCO-IQ3_S-00001-of-00002.gguf + presence_penalty: 0 + repeat_penalty: 1 + temperature: 1 + top_k: 20 + top_p: 0.95 + template: + use_tokenizer_template: true - kind: model id: qwen3.8-flash-next-gsq-rco-q2-0 name: qwen3.8-flash-next-gsq-rco-q2-0