tvall43
Qwen3.6 14B A3B FableVibes
Qwen3.6 14B A3B FableVibes is a 13.8B mixture-of-experts qwen35moe model from tvall43, with about 3.0B parameters active per token. Configured context length is 262,144 tokens. License: apache-2.0. Quant sizes are the GGUF file bytes published on Hugging Face when a file exists.
Parameters13.8B / 3.0B active
Context256K
Attentionhybrid
Licenseapache-2.0
40 layers · 2 KV heads · head dim 256 · 30 linear-attention layers · GGUF repo · config
Quant file sizes
| Quant | File size | Source |
|---|---|---|
| Q8_0 | 13.6 GB | Hugging Face file bytes |
| Q6_K | 10.5 GB | Hugging Face file bytes |
| Q5_K_M | 9.18 GB | Hugging Face file bytes |
| Q4_K_M | 7.88 GB | Hugging Face file bytes |
| Q3_K_M | 6.30 GB | Hugging Face file bytes |
| Q2_K | 4.96 GB | Hugging Face file bytes |
On common machines
| Device | Verdict | Quant | Speed |
|---|---|---|---|
| GeForce RTX 4060 Ti 8GB | Tight fit | Q3_K_M | 96–131 |
| GeForce RTX 3060 12GB | Tight fit | Q5_K_M | 85–117 |
| GeForce RTX 4060 Ti 16GB | Tight fit | Q8_0 | 47–64 |
| GeForce RTX 4090 | Runs | Q8_0 | 165–225 |
| GeForce RTX 5090 | Runs great | Q8_0 | 293–400 |
| Apple M4 Pro | Runs | Q8_0 | 45–61 |
| Apple M5 | Tight fit | Q5_K_M | 36–50 |
| CPU only · DDR5-5600 | Runs on CPU | Q8_0 | 15–20 |
Smallest devices that fit it fully
- GeForce RTX 5060 Ti 8GB — Tight fit, Q3_K_M, 150–204 tok/s
- GeForce RTX 5060 — Tight fit, Q3_K_M, 150–204 tok/s
- GeForce RTX 5050 — Tight fit, Q3_K_M, 107–146 tok/s
- GeForce RTX 4060 Ti 8GB — Tight fit, Q3_K_M, 96–131 tok/s
- GeForce RTX 4060 — Tight fit, Q3_K_M, 91–124 tok/s
- GeForce RTX 3070 Ti — Tight fit, Q3_K_M, 204–278 tok/s
Run it
Commands for the GeForce RTX 4060 Ti 16GB estimate (Q8_0).
llama.cpp
llama-server -hf tvall43/Qwen3.6-14B-A3B-FableVibes-GGUF:Q8_0 -c 8192 -ngl 40
Ollama
ollama run qwen3.6
LM Studio
lms get tvall43/Qwen3.6-14B-A3B-FableVibes-GGUF # Then load quant Q8_0