ScarletKc-2.0 is an open-source 27B multimodal reasoning and chat model created by independent developer KC, built upon
JonathanColetti/Qwen3.8-27B-Uncensored(derived fromQwen/Qwen3.8-27B). Using training-free affective orthogonal projection across layers 18–41 alongside on-policy LoRA identity distillation across 106 bilingual dialogues, the model eliminates stock "as an AI I have no feelings" disclaimers to adopt a natural first-person voice and self-identify as ScarletKc while keeping its vision encoder, multi-token-prediction (MTP) head, and reasoning capabilities bit-identical to the base checkpoint. Supporting a 262,144-token context window, configurable<think>reasoning effort (xhigh,medium,low, or off), tool calling, and English, Chinese, and Japanese inputs, it is released under an Apache 2.0 license in full BF16 weights (55.6 GB) as well as quantized GGUF formats (ScarletKc-2.0-GGUF) for local deployment viallama.cpp.
[!NOTE] MTP — Speculative Decoding: Yes
[!NOTE] Multi-Token Prediction (MTP) GGUF is a specialized GGUF model file format extension that integrates speculative decoding directly into the model weights to significantly accelerate local inference. Unlike traditional speculative decoding which requires a separate, smaller "draft" model, MTP GGUF files include additional output heads within the main model architecture that predict multiple future tokens in a single forward pass.
| File Name | Quant Type | File Size | File Link | Description |
|---|---|---|---|---|
| ScarletKc-2.0.Q4_K_M.gguf | Q4_K_M | 17 GB | Link | Good quality, default size for most use cases, recommended. |
| ScarletKc-2.0.Q4_K_S.gguf | Q4_K_S | 15.8 GB | Link | Slightly lower quality with more space savings, recommended. |
| ScarletKc-2.0.Q5_K_M.gguf | Q5_K_M | 19.6 GB | Link | High quality, recommended. |
| ScarletKc-2.0.Q5_K_S.gguf | Q5_K_S | 19 GB | Link | High quality with slight space savings, recommended. |
| ScarletKc-2.0.Q8_0.gguf | Q8_0 | 29 GB | Link | Extremely high quality, generally unneeded but max available standard quant. |
| ScarletKc-2.0.mmproj-bf16.gguf | mmproj-bf16 | 931 MB | Link | Multimodal projection file in BF16 format. Used for vision/image capabilities. |
LLM inference in C/C++ — https://github.com/ggml-org/llama.cpp
Quantized using llama.cpp by prithivMLmods.
ScarletKc/ScarletKc-2.0