Instructions to use m9ter1a/LFM2.5-24B-A2B-GGUF with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Notebooks
Google Colab
Kaggle
Local Apps Settings

How to use m9ter1a/LFM2.5-24B-A2B-GGUF with llama.cpp:

Install (macOS, Linux)

curl -LsSf https://llama.app/install.sh | sh
# Start a local OpenAI-compatible server with a web UI:
llama serve -hf m9ter1a/LFM2.5-24B-A2B-GGUF:Q4_K_M
# Run inference directly in the terminal:
llama cli -hf m9ter1a/LFM2.5-24B-A2B-GGUF:Q4_K_M

Install from WinGet (Windows)

winget install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama serve -hf m9ter1a/LFM2.5-24B-A2B-GGUF:Q4_K_M
# Run inference directly in the terminal:
llama cli -hf m9ter1a/LFM2.5-24B-A2B-GGUF:Q4_K_M

Build from source code

git clone https://github.com/ggml-org/llama.cpp.git
cd llama.cpp
cmake -B build
cmake --build build -j --target llama-app
# Start a local OpenAI-compatible server with a web UI:
./build/bin/llama serve -hf m9ter1a/LFM2.5-24B-A2B-GGUF:Q4_K_M
# Run inference directly in the terminal:
./build/bin/llama cli -hf m9ter1a/LFM2.5-24B-A2B-GGUF:Q4_K_M

Use Docker

docker model run hf.co/m9ter1a/LFM2.5-24B-A2B-GGUF:Q4_K_M

Llama new
LM Studio
Jan

vLLM

How to use m9ter1a/LFM2.5-24B-A2B-GGUF with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "m9ter1a/LFM2.5-24B-A2B-GGUF"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "m9ter1a/LFM2.5-24B-A2B-GGUF",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/m9ter1a/LFM2.5-24B-A2B-GGUF:Q4_K_M

Ollama
How to use m9ter1a/LFM2.5-24B-A2B-GGUF with Ollama:
```
ollama run hf.co/m9ter1a/LFM2.5-24B-A2B-GGUF:Q4_K_M
```
Unsloth Desktop

How to use m9ter1a/LFM2.5-24B-A2B-GGUF with Pi:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf m9ter1a/LFM2.5-24B-A2B-GGUF:Q4_K_M

Configure the model in Pi

# Install Pi:
npm install -g @earendil-works/pi-coding-agent
# Add to ~/.pi/agent/models.json:
{
  "providers": {
    "llama-cpp": {
      "baseUrl": "http://localhost:8080/v1",
      "api": "openai-completions",
      "apiKey": "none",
      "models": [
        {
          "id": "m9ter1a/LFM2.5-24B-A2B-GGUF:Q4_K_M"
        }
      ]
    }
  }
}

Run Pi

# Start Pi in your project directory:
pi

Docker Model Runner
How to use m9ter1a/LFM2.5-24B-A2B-GGUF with Docker Model Runner:
```
docker model run hf.co/m9ter1a/LFM2.5-24B-A2B-GGUF:Q4_K_M
```

Lemonade

How to use m9ter1a/LFM2.5-24B-A2B-GGUF with Lemonade:

Pull the model

# Download Lemonade from https://lemonade-server.ai/
lemonade pull m9ter1a/LFM2.5-24B-A2B-GGUF:Q4_K_M

Run and chat with the model

lemonade run user.LFM2.5-24B-A2B-GGUF-Q4_K_M

List all available models

lemonade list

Hermes Agent

How to use m9ter1a/LFM2.5-24B-A2B-GGUF with Hermes Agent:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf m9ter1a/LFM2.5-24B-A2B-GGUF:Q4_K_M

Configure Hermes

# Install Hermes:
curl -fsSL https://hermes-agent.nousresearch.com/install.sh | bash
hermes setup
# Point Hermes at the local server:
hermes config set model.provider custom
hermes config set model.base_url http://127.0.0.1:8080/v1
hermes config set model.default m9ter1a/LFM2.5-24B-A2B-GGUF:Q4_K_M

Run Hermes

hermes

Atomic Chat

OpenClaw

How to use m9ter1a/LFM2.5-24B-A2B-GGUF with OpenClaw:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf m9ter1a/LFM2.5-24B-A2B-GGUF:Q4_K_M

Configure OpenClaw

# Install OpenClaw:
npm install -g openclaw@latest
# Register the local server and set it as the default model:
openclaw onboard --non-interactive --mode local \
  --auth-choice custom-api-key \
  --custom-base-url http://127.0.0.1:8080/v1 \
  --custom-model-id "m9ter1a/LFM2.5-24B-A2B-GGUF:Q4_K_M" \
  --custom-provider-id llama-cpp \
  --custom-compatibility openai \
  --custom-text-input \
  --accept-risk \
  --skip-health

Run OpenClaw

openclaw agent --local --agent main --message "Hello from Hugging Face"

LFM2.5-24B-A2B-GGUF / comparison.csv

m9ter1a

Add model card, chart, comparison data, scripts and raw results

e95e6f6 verified 27 days ago

Raw History Blame Contribute Delete

3.08 kB

publisher,quant,is_ours,hf_repo,file,size_bytes,size_gb,size_gib,bits_per_weight,pct_of_bf16_size,kld_mean,kld_mean_stderr,kld_median,kld_p99,top1_same_pct,top1_same_stderr_pct,bench_items,toolcall_same_as_bf16_pct,toolcall_same_as_bf16_ci95,toolcall_same_decision_pct,bfcl_correct_pct,bfcl_correct_ci95,subset178_toolcall_same_as_bf16_pct,subset178_bfcl_correct_pct,bfcl_simple_pct,bfcl_multiple_pct,bfcl_parallel_pct,bfcl_parallel_multiple_pct,bfcl_live_simple_pct,bfcl_live_multiple_pct,bfcl_irrelevance_pct,bfcl_live_irrelevance_pct,note
reference,BF16,False,LiquidAI/LFM2.5-24B-A2B,converted to GGUF from the original safetensors,47962095648,47.96,44.67,16.01,100.0,0.0,0.0,0.0,0.0,100.0,0.0,300,100.0,0.0,100.0,72.7,5.0,100.0,75.3,76.0,74.0,82.5,65.0,72.5,72.5,75.0,55.0,the reference every other row is measured against
m9ter1a (ours),Q5_K_M,True,m9ter1a/LFM2.5-24B-A2B-GGUF,LFM2.5-24B-A2B-Q5_K_M.gguf,17368318176,17.37,16.18,5.8,36.2,0.038862,0.000447,0.017617,0.333415,91.288,0.139,178,82.6,5.6,100.0,77.0,6.2,82.6,77.0,78.0,78.0,77.5,73.7,,,,,"tool-call bench covers 178 of 300 items (simple, multiple, parallel, parallel_multiple); use the subset178 columns to compare"
AtomicChat,AD-Q4_K_M,False,AtomicChat/LFM2.5-24B-A2B-GGUF,LFM2.5-24B-A2B-AD-Q4_K_M.gguf,15359770880,15.36,14.3,5.13,32.0,0.058737,0.000637,0.027224,0.502832,89.396,0.152,300,81.0,4.4,98.0,73.7,5.0,79.8,75.8,80.0,72.0,75.0,75.0,80.0,72.5,70.0,50.0,
m9ter1a (ours),Q4_K_M,True,m9ter1a/LFM2.5-24B-A2B-GGUF,LFM2.5-24B-A2B-Q4_K_M.gguf,14952399072,14.95,13.93,4.99,31.2,0.065806,0.000754,0.030597,0.54583,88.717,0.156,300,81.0,4.4,98.0,78.0,4.7,82.0,78.1,76.0,76.0,80.0,80.0,85.0,77.5,80.0,65.0,
unsloth,UD-Q4_K_XL,False,unsloth/LFM2.5-24B-A2B-Preview-GGUF,LFM2.5-24B-A2B-Preview-UD-Q4_K_XL.gguf,14876901632,14.88,13.86,4.96,31.0,0.061665,0.000646,0.028748,0.511352,89.079,0.154,300,82.0,4.3,97.7,73.0,5.0,82.6,74.7,76.0,76.0,77.5,67.5,75.0,72.5,80.0,50.0,
LiquidAI (official),Q4_K_M,False,LiquidAI/LFM2.5-24B-A2B-GGUF,LFM2.5-24B-A2B-Q4_K_M.gguf,14526258464,14.53,13.53,4.85,30.3,0.132477,0.001209,0.067449,1.024848,84.062,0.181,300,77.0,4.8,97.0,77.0,4.8,78.1,78.1,88.0,76.0,72.5,72.5,82.5,75.0,75.0,65.0,
AtomicChat,AD-IQ4_XS,False,AtomicChat/LFM2.5-24B-A2B-GGUF,LFM2.5-24B-A2B-AD-IQ4_XS.gguf,14496268544,14.5,13.5,4.84,30.2,0.068324,0.000849,0.030282,0.607643,88.487,0.158,300,80.3,4.5,98.0,73.0,5.0,81.5,74.7,76.0,76.0,67.5,75.0,75.0,75.0,70.0,60.0,
LiquidAI (official),QAD-Q4_0,False,LiquidAI/LFM2.5-24B-A2B-GGUF,LFM2.5-24B-A2B-QAD-Q4_0.gguf,13578190880,13.58,12.65,4.53,28.3,0.212861,0.001703,0.111898,1.619704,80.242,0.197,300,75.3,4.9,98.3,67.0,5.3,73.6,63.5,66.0,62.0,60.0,62.5,75.0,75.0,75.0,65.0,"weights retrained (quantization-aware distillation), so it deliberately differs from BF16: KLD and top-1 measure that distance, not quality"
m9ter1a (ours),IQ4_XS,True,m9ter1a/LFM2.5-24B-A2B-GGUF,LFM2.5-24B-A2B-IQ4_XS.gguf,13300891872,13.3,12.39,4.44,27.7,0.074496,0.000723,0.035426,0.617043,87.818,0.162,300,77.0,4.8,99.0,75.7,4.9,73.6,78.1,84.0,76.0,77.5,70.0,82.5,75.0,70.0,55.0,

	publisher,quant,is_ours,hf_repo,file,size_bytes,size_gb,size_gib,bits_per_weight,pct_of_bf16_size,kld_mean,kld_mean_stderr,kld_median,kld_p99,top1_same_pct,top1_same_stderr_pct,bench_items,toolcall_same_as_bf16_pct,toolcall_same_as_bf16_ci95,toolcall_same_decision_pct,bfcl_correct_pct,bfcl_correct_ci95,subset178_toolcall_same_as_bf16_pct,subset178_bfcl_correct_pct,bfcl_simple_pct,bfcl_multiple_pct,bfcl_parallel_pct,bfcl_parallel_multiple_pct,bfcl_live_simple_pct,bfcl_live_multiple_pct,bfcl_irrelevance_pct,bfcl_live_irrelevance_pct,note
	reference,BF16,False,LiquidAI/LFM2.5-24B-A2B,converted to GGUF from the original safetensors,47962095648,47.96,44.67,16.01,100.0,0.0,0.0,0.0,0.0,100.0,0.0,300,100.0,0.0,100.0,72.7,5.0,100.0,75.3,76.0,74.0,82.5,65.0,72.5,72.5,75.0,55.0,the reference every other row is measured against
	m9ter1a (ours),Q5_K_M,True,m9ter1a/LFM2.5-24B-A2B-GGUF,LFM2.5-24B-A2B-Q5_K_M.gguf,17368318176,17.37,16.18,5.8,36.2,0.038862,0.000447,0.017617,0.333415,91.288,0.139,178,82.6,5.6,100.0,77.0,6.2,82.6,77.0,78.0,78.0,77.5,73.7,,,,,"tool-call bench covers 178 of 300 items (simple, multiple, parallel, parallel_multiple); use the subset178 columns to compare"
	AtomicChat,AD-Q4_K_M,False,AtomicChat/LFM2.5-24B-A2B-GGUF,LFM2.5-24B-A2B-AD-Q4_K_M.gguf,15359770880,15.36,14.3,5.13,32.0,0.058737,0.000637,0.027224,0.502832,89.396,0.152,300,81.0,4.4,98.0,73.7,5.0,79.8,75.8,80.0,72.0,75.0,75.0,80.0,72.5,70.0,50.0,
	m9ter1a (ours),Q4_K_M,True,m9ter1a/LFM2.5-24B-A2B-GGUF,LFM2.5-24B-A2B-Q4_K_M.gguf,14952399072,14.95,13.93,4.99,31.2,0.065806,0.000754,0.030597,0.54583,88.717,0.156,300,81.0,4.4,98.0,78.0,4.7,82.0,78.1,76.0,76.0,80.0,80.0,85.0,77.5,80.0,65.0,
	unsloth,UD-Q4_K_XL,False,unsloth/LFM2.5-24B-A2B-Preview-GGUF,LFM2.5-24B-A2B-Preview-UD-Q4_K_XL.gguf,14876901632,14.88,13.86,4.96,31.0,0.061665,0.000646,0.028748,0.511352,89.079,0.154,300,82.0,4.3,97.7,73.0,5.0,82.6,74.7,76.0,76.0,77.5,67.5,75.0,72.5,80.0,50.0,
	LiquidAI (official),Q4_K_M,False,LiquidAI/LFM2.5-24B-A2B-GGUF,LFM2.5-24B-A2B-Q4_K_M.gguf,14526258464,14.53,13.53,4.85,30.3,0.132477,0.001209,0.067449,1.024848,84.062,0.181,300,77.0,4.8,97.0,77.0,4.8,78.1,78.1,88.0,76.0,72.5,72.5,82.5,75.0,75.0,65.0,
	AtomicChat,AD-IQ4_XS,False,AtomicChat/LFM2.5-24B-A2B-GGUF,LFM2.5-24B-A2B-AD-IQ4_XS.gguf,14496268544,14.5,13.5,4.84,30.2,0.068324,0.000849,0.030282,0.607643,88.487,0.158,300,80.3,4.5,98.0,73.0,5.0,81.5,74.7,76.0,76.0,67.5,75.0,75.0,75.0,70.0,60.0,
	LiquidAI (official),QAD-Q4_0,False,LiquidAI/LFM2.5-24B-A2B-GGUF,LFM2.5-24B-A2B-QAD-Q4_0.gguf,13578190880,13.58,12.65,4.53,28.3,0.212861,0.001703,0.111898,1.619704,80.242,0.197,300,75.3,4.9,98.3,67.0,5.3,73.6,63.5,66.0,62.0,60.0,62.5,75.0,75.0,75.0,65.0,"weights retrained (quantization-aware distillation), so it deliberately differs from BF16: KLD and top-1 measure that distance, not quality"
	m9ter1a (ours),IQ4_XS,True,m9ter1a/LFM2.5-24B-A2B-GGUF,LFM2.5-24B-A2B-IQ4_XS.gguf,13300891872,13.3,12.39,4.44,27.7,0.074496,0.000723,0.035426,0.617043,87.818,0.162,300,77.0,4.8,99.0,75.7,4.9,73.6,78.1,84.0,76.0,77.5,70.0,82.5,75.0,70.0,55.0,