ajgazin
Swift-Qwen3.8-27B-Uncensored-Dynamic-MTP
Image input
- Architecture
- clip
- License
- Conditional
- Base model
- ajgazin/Swift-Qwen3.8-27B-Uncensored-MTP
- License
- swift-open-license-1.0
- llama.cpp
- b6768+
About Swift-Qwen3.8-27B-Uncensored-Dynamic-MTP
Swift-Qwen3.8-27B-Uncensored-Dynamic-MTP is a GGUF model based on ajgazin/Swift-Qwen3.8-27B-Uncensored-MTP on Hugging Face. GGUF files are published by ajgazin/Swift-Qwen3.8-27B-Uncensored-Dynamic-MTP-GGUF. It has the clip architecture (values written in the GGUF).
There are 12 quantizations, with file sizes from 9.83 GB (UD-Q2_K_XL) to 54.7 GB (BF16) (actual sizes on Hugging Face).
Running UD-Q2_K_XL (9.83 GB) with an 8,192-token context needs an estimated 11.6 GB of memory, and an estimated 13.2 GB at 32,768 tokens. For 8,192 tokens, a GPU with 12 GB or more of VRAM (such as GeForce RTX 3080 Ti, GeForce RTX 5070), or a Mac with 12 GB or more of unified memory, is a rough guide (macOS caps how much memory the GPU can use, so leave some headroom).
The original model is licensed under Swift Open License v1.0, which allows commercial use with conditions such as scale or purpose. Check the original license for the terms of use. An mmproj file for image input is also available.
File size and estimated memory per quantization
- Up to 8 GB
- Up to 16 GB
- Up to 24 GB
- Up to 48 GB
- Up to 128 GB
- More
File sizes are the actual values on Hugging Face. Memory is estimated for each context length (KV cache f16). The formula is in How memory is calculated.
All memory figures are Estimated (KV cache f16). It should run on a setup with at least this much VRAM (or Mac unified memory). Click a value to see the devices and models that fit that amount (Find by GPU / Mac).
Using image input: This model supports image input. To use images, an mmproj file is loaded as well.
About the "quality guide" (how far a quantization drifts from the original)
For each quantization type, how far the output drifts from the original model, grouped into six levels from values llama.cpp measured on Llama-3-8B.
The value is mean KLD (the difference between the original and quantized model in next-token probabilities); smaller means closer to the original. Quantizations made with an imatrix (importance data) come out slightly smaller. Levels use the value without an imatrix (or with one, for types that only have that).
These values were not measured on this model. Larger models and MoE models tend to drift less with the same type. Distributor-specific mixed types (such as UD-Q4_K_XL) cannot be matched by type name alone, so they show "-".
| Quantization | Quality guide | KLD (no imatrix) | KLD (imatrix) | Size (Llama-3-8B) |
|---|---|---|---|---|
| Q8_0 | Nearly identical | 0.0014 | - | 7.96 GiB |
| Q6_K | Very small | 0.0055 | - | 6.14 GiB |
| Q5_K_M | Very small | 0.0108 | - | 5.33 GiB |
| Q5_K_S | Small | 0.0166 | - | 5.21 GiB |
| Q5_1 | Small | 0.0180 | - | 5.65 GiB |
| Q5_0 | Small | 0.0222 | - | 5.21 GiB |
| Q4_K_M | Small | 0.0313 | 0.0282 | 4.58 GiB |
| Q4_K_S | Medium | 0.0431 | 0.0320 | 4.37 GiB |
| IQ4_NL | Small | - | 0.0357 | 4.35 GiB |
| IQ4_XS | Small | - | 0.0363 | 4.14 GiB |
| Q4_1 | Medium | 0.0717 | - | 4.78 GiB |
| Q4_0 | Medium | 0.0719 | - | 4.34 GiB |
| Q3_K_L | Medium | 0.0872 | 0.0731 | 4.03 GiB |
| Q3_K_M | Medium | 0.1019 | 0.0844 | 3.74 GiB |
| IQ3_M | Medium | - | 0.1025 | 3.53 GiB |
| IQ3_S | Medium | - | 0.1113 | 3.42 GiB |
| IQ3_XS | Large | - | 0.1387 | 3.28 GiB |
| IQ3_XXS | Large | - | 0.1836 | 3.05 GiB |
| Q3_K_S | Large | 0.2282 | 0.1998 | 3.41 GiB |
| IQ2_M | Large | - | 0.3260 | 2.74 GiB |
| Q2_K | Very large | 0.4451 | 0.3322 | 2.96 GiB |
| Q2_K_S | Very large | - | 0.4034 | 2.96 GiB |
| IQ2_S | Very large | - | 0.4392 | 2.56 GiB |
| IQ2_XS | Very large | - | 0.5463 | 2.43 GiB |
| IQ2_XXS | Very large | - | 0.8120 | 2.24 GiB |
| IQ1_M | Very large | - | 1.3931 | 2.01 GiB |
| IQ1_S | Very large | - | 2.2546 | 1.88 GiB |
Source: llama.cpp (MIT), measured on Llama-3-8B against FP16, Wikitext-2, revision f364eb6f. Original table · Reading and choosing quantizations
GGUF source
ajgazin/Swift-Qwen3.8-27B-Uncensored-Dynamic-MTP-GGUF
Published by the original developer Modified version (name includes abliterated etc.) About this label
Last checked: Oct 19, 2026
- Architecture
- clip
- llama.cpp b6768 and later
- License
- swift-open-license-1.0 Conditional
The mmproj file for image input (BF16) is 0.93 GB Measured . It is not included in the memory above, so add it if you use images.
Ollama ollama run hf.co/ajgazin/Swift-Qwen3.8-27B-Uncensored-Dynamic-MTP-GGUF:UD-Q2_K_XL
llama.cpp llama-server -hf ajgazin/Swift-Qwen3.8-27B-Uncensored-Dynamic-MTP-GGUF:UD-Q2_K_XL
Generation speed on common GPUs and Macs (estimated)
A rough guide to how fast UD-Q2_K_XL (9.83 GB) generates text. Each token reads the model weights once, so the upper limit is memory bandwidth ÷ file size. The guide is 60–80% of that, where most published measurements fall.
| Device | Memory bandwidth | Guide (tokens/s) Estimated | Upper limit |
|---|---|---|---|
| Radeon RX 7600 XT 16 GB | 288 GB/s | 18–23 | 29 |
| Radeon RX 6950 XT 16 GB | 576 GB/s | 35–47 | 59 |
| GeForce RTX 5090 32 GB | 1,792 GB/s | 109–146 | 182 |
| Apple M2 16 GB | 100 GB/s | 6.1–8.1 | 10 |
| Apple M2 Pro 16 GB | 200 GB/s | 12–16 | 20 |
| Apple M5 Ultra 96 GB | 1,200 GB/s | 73–98 | 122 |
Estimates only. Prompt processing speed (limited by compute) is not included. Longer contexts and offloading part of the model to the CPU make it slower. Macs are shown only when installed memory is at least 4/3 of what is needed. Devices without published bandwidth (some GeForce cards, etc.) are not shown.
License and commercial use
Swift Open License v1.0 Conditional
Commercial use by organizations with total revenue of USD 1 million or more in the most recent fiscal year is not permitted under this license (a separate agreement is needed). Below that, commercial use is allowed.
Based on the license on the original model's Hugging Face page. Original license text (checked Oct 18, 2026)
The summary states facts this site read from the original text and is not legal advice. Always check the original license and its latest terms before use. All licenses
How to run
Mainline llama.cpp can load this model's format (clip) from b6768 (Oct 15, 2025) onward. Ollama and LM Studio bundle llama.cpp internally, so an older app fails to load it with errors such as "unknown model architecture". If that happens, update the app.
Example commands using UD-Q2_K_XL. To use another quantization, replace the name with one from the table above. How to run local LLMs
Ollama
Run it by pointing directly at the GGUF on Hugging Face.
ollama run hf.co/ajgazin/Swift-Qwen3.8-27B-Uncensored-Dynamic-MTP-GGUF:UD-Q2_K_XL
llama.cpp
Pass the repository and quantization name to llama-server (or llama-cli) with -hf.
llama-server -hf ajgazin/Swift-Qwen3.8-27B-Uncensored-Dynamic-MTP-GGUF:UD-Q2_K_XL
LM Studio
Paste ajgazin/Swift-Qwen3.8-27B-Uncensored-Dynamic-MTP-GGUF into the search box, download it, and choose the quantization from the file list.
Other models from ajgazin
Frequently asked questions
- How much memory do I need to run Swift-Qwen3.8-27B-Uncensored-Dynamic-MTP (UD-Q2_K_XL)?
- The file size is 9.83 GB (measured). With a 8,192-token context, the estimated memory including the KV cache is 11.6 GB. The formula is published at https://localllm.az-arte.com/en/guide/memory-calculation.
- Can Swift-Qwen3.8-27B-Uncensored-Dynamic-MTP be used commercially?
- The original model's license is Swift Open License v1.0 (Commercial use with conditions).Commercial use by organizations with total revenue of USD 1 million or more in the most recent fiscal year is not permitted under this license (a separate agreement is needed). Below that, commercial use is allowed. Check the original license text before deciding.
- Does Swift-Qwen3.8-27B-Uncensored-Dynamic-MTP run in 8 GB, 12 GB, 16 GB or 24 GB of VRAM?
- 8 GB: no quantization fits; 12 GB: up to UD-IQ3_XXS (est. 12.7 GB); 16 GB: up to UD-Q4_K_S (est. 17.1 GB); 24 GB: up to UD-Q5_K_M (est. 21.5 GB) (estimated at 8,192 tokens with an f16 KV cache, files from the most-downloaded source).
- What is the command to run Swift-Qwen3.8-27B-Uncensored-Dynamic-MTP with Ollama?
- Run "ollama run hf.co/ajgazin/Swift-Qwen3.8-27B-Uncensored-Dynamic-MTP-GGUF:UD-Q2_K_XL" to use the GGUF on Hugging Face directly. To use another quantization, replace the name at the end.