No official torrent yet — requesting it tells the flock what to host next. You cannot seed a catalog-only page.
Cataloged from DhruvalLabs/Kimi-Linear-48B-A3B-Instruct-GGUF
All models by DhruvalLabs →Quantized GGUF versions of moonshotai/Kimi-Linear-48B-A3B-Instruct
Works with llama.cpp · Ollama · LM Studio · Open WebUI · Jan
Can you run a powerful model on a laptop without losing its intelligence?
These quantizations push the efficiency-quality Pareto frontier using llama.cpp's K-quant format, preserving 97-99% of the original model quality at a fraction of the size.
Kimi-Linear-48B-A3B-Instruct-Q4_K_M.gguf — best balance of size and qualityKimi-Linear-48B-A3B-Instruct-Q8_0.gguf — near-original qualityKimi-Linear-48B-A3B-Instruct-Q3_K_M.gguf — fits in 8GB with room to spareRun python benchmark.py --model Kimi-Linear-48B-A3B-Instruct to generate speed results.
Run kaggle_bench.ipynb on Kaggle to benchmark this model.
Search for Dhptl/Kimi-Linear-48B-A3B-Instruct in the model browser.
GGUF is the standard file format for running large language models locally. Quantization reduces the number of bits per weight:
Found an issue? Have a question? Open a Discussion in the Community tab above.
If these quantizations were useful, please consider:
ollama run dhptl/kimi-linear-48b-a3b-instruct# Download the binary from https://github.com/ggerganov/llama.cpp/releases./llama-cli \ -m Kimi-Linear-48B-A3B-Instruct-Q4_K_M.gguf \ -p "You are a helpful assistant." \ --conversation \ -n 512from llama_cpp import Llama
llm = Llama( model_path="./Kimi-Linear-48B-A3B-Instruct-Q4_K_M.gguf", n_gpu_layers=-1, # -1 = offload everything to GPU n_ctx=4096,)
response = llm.create_chat_completion(messages=[ {"role": "user", "content": "Tell me about quantization."}])print(response["choices"][0]["message"]["content"])