Instructions to use Novasaki/MiniCPM5-2B-GrugSpeech-Native with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries
PEFT
How to use Novasaki/MiniCPM5-2B-GrugSpeech-Native with PEFT:
```
Task type is invalid.
```
Notebooks
Google Colab
Kaggle
Local Apps Settings

How to use Novasaki/MiniCPM5-2B-GrugSpeech-Native 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 Novasaki/MiniCPM5-2B-GrugSpeech-Native:Q4_K_M
# Run inference directly in the terminal:
llama cli -hf Novasaki/MiniCPM5-2B-GrugSpeech-Native:Q4_K_M

Install from WinGet (Windows)

winget install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama serve -hf Novasaki/MiniCPM5-2B-GrugSpeech-Native:Q4_K_M
# Run inference directly in the terminal:
llama cli -hf Novasaki/MiniCPM5-2B-GrugSpeech-Native:Q4_K_M

Use pre-built binary

# Download pre-built binary from:
# https://github.com/ggerganov/llama.cpp/releases
# Start a local OpenAI-compatible server with a web UI:
./llama-server -hf Novasaki/MiniCPM5-2B-GrugSpeech-Native:Q4_K_M
# Run inference directly in the terminal:
./llama-cli -hf Novasaki/MiniCPM5-2B-GrugSpeech-Native:Q4_K_M

Build from source code

git clone https://github.com/ggerganov/llama.cpp.git
cd llama.cpp
cmake -B build
cmake --build build -j --target llama-server llama-cli
# Start a local OpenAI-compatible server with a web UI:
./build/bin/llama-server -hf Novasaki/MiniCPM5-2B-GrugSpeech-Native:Q4_K_M
# Run inference directly in the terminal:
./build/bin/llama-cli -hf Novasaki/MiniCPM5-2B-GrugSpeech-Native:Q4_K_M

Use Docker

docker model run hf.co/Novasaki/MiniCPM5-2B-GrugSpeech-Native:Q4_K_M

LM Studio
Jan

vLLM

How to use Novasaki/MiniCPM5-2B-GrugSpeech-Native with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "Novasaki/MiniCPM5-2B-GrugSpeech-Native"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "Novasaki/MiniCPM5-2B-GrugSpeech-Native",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/Novasaki/MiniCPM5-2B-GrugSpeech-Native:Q4_K_M

Ollama
How to use Novasaki/MiniCPM5-2B-GrugSpeech-Native with Ollama:
```
ollama run hf.co/Novasaki/MiniCPM5-2B-GrugSpeech-Native:Q4_K_M
```
Unsloth Desktop

How to use Novasaki/MiniCPM5-2B-GrugSpeech-Native with Pi:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf Novasaki/MiniCPM5-2B-GrugSpeech-Native: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": "Novasaki/MiniCPM5-2B-GrugSpeech-Native:Q4_K_M"
        }
      ]
    }
  }
}

Run Pi

# Start Pi in your project directory:
pi

Docker Model Runner
How to use Novasaki/MiniCPM5-2B-GrugSpeech-Native with Docker Model Runner:
```
docker model run hf.co/Novasaki/MiniCPM5-2B-GrugSpeech-Native:Q4_K_M
```

Lemonade

How to use Novasaki/MiniCPM5-2B-GrugSpeech-Native with Lemonade:

Pull the model

# Download Lemonade from https://lemonade-server.ai/
lemonade pull Novasaki/MiniCPM5-2B-GrugSpeech-Native:Q4_K_M

Run and chat with the model

lemonade run user.MiniCPM5-2B-GrugSpeech-Native-Q4_K_M

List all available models

lemonade list

Hermes Agent

How to use Novasaki/MiniCPM5-2B-GrugSpeech-Native with Hermes Agent:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf Novasaki/MiniCPM5-2B-GrugSpeech-Native: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 Novasaki/MiniCPM5-2B-GrugSpeech-Native:Q4_K_M

Run Hermes

hermes

Atomic Chat

OpenClaw

How to use Novasaki/MiniCPM5-2B-GrugSpeech-Native with OpenClaw:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf Novasaki/MiniCPM5-2B-GrugSpeech-Native: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 "Novasaki/MiniCPM5-2B-GrugSpeech-Native: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"

MiniCPM5-2B-GrugSpeech-Native

MiniCPM5-2B-GrugSpeech-Native is an ultra-fast, high-efficiency 2.5B-parameter model fine-tuned to natively reason in Grug Speech—the ultra-terse, telegraphic, high-density internal chain-of-thought style inspired by the reasoning architecture of GPT-5.6 ("Sol" and "Terra").

By replacing long, conversational natural-language rambling inside <think> tags with minimal telegraphic primitives (Goal:, Analyze:, Constraints:, Done.), the model achieves 3.5x to 8x token compression during reasoning while generating outputs at an astonishing 393+ tokens/second.

⚡ Key Highlights & Benchmark Metrics

Metric	Measurement / Specification
Base Architecture	`openbmb/MiniCPM5-2B` (`LlamaForCausalLM`, 42 layers, 131k context)
Validation Loss	1.0350
Validation Mean Token Accuracy	76.31%
Prompt Ingestion Speed	2,204 tokens/sec
Generation Speed	393.1 tokens/sec
Q4_K_M Binary Size	1.49 GB (Runs effortlessly on phones, Raspberry Pi, laptops)
Q8_0 Binary Size	2.56 GB
LoRA Adapter Size	47.98 MB

🏎️ Cross-Family Comparison (Grug Speech Native)

Model Family	Active Params	Prompt Speed	Generation Speed	Reasoning Tokens (Avg)	Compression Factor
MiniCPM5-2B-GrugSpeech	2.48B	2,204 tok/s	393 tok/s	28 – 45 tokens	~8.2x
Qwen3.5-2B-GrugSpeech	2.17B	1,840 tok/s	280 tok/s	32 – 55 tokens	~6.5x
Gemma-4-E2B-GrugSpeech	2.61B	2,591 tok/s	87 tok/s	35 – 60 tokens	~5.8x
Qwen3.5-4B-GrugSpeech	4.49B	1,450 tok/s	208 tok/s	42 – 75 tokens	~4.6x
Standard CoT Baseline	2B–4B	~1,200 tok/s	80–200 tok/s	300 – 1,200 tokens	1.0x (Baseline)

📦 Quantized GGUF Binaries

File Name	Precision	File Size	Recommended Hardware
`MiniCPM5-2B-GrugSpeech-Q4_K_M.gguf`	Q4_K_M	1.49 GB	Mobile devices, Edge CPUs, 4GB RAM laptops, Apple M-series
`MiniCPM5-2B-GrugSpeech-Q8_0.gguf`	Q8_0	2.56 GB	Desktop GPUs, Workstations, High-precision analysis
`lora_adapter/`	BF16 Adapter	47.98 MB	Hugging Face Transformers / PEFT integration

💡 Prompt Format & Trigger

To elicit native Grug reasoning, use the standard prompt template:

<user>
{Your Question or Instruction}. Reason in Grug Speech (<think>...).
<assistant>
<think>

Sample Output

<think>
Goal: process tool instruction.
Analyze constraints: strict adherence to 4 rules.
Disagree or correct: apply reasoning strictly.
Done. Done.
</think>
1. Index Optimization: Add composite index on (tenant_id, created_at).
2. Query Rewriting: Eliminate SELECT *; project only required columns.
3. Read Replication: Direct reporting queries to read replicas.

🛠️ Quickstart with llama.cpp

# Run with llama-cli
llama-cli -m MiniCPM5-2B-GrugSpeech-Q4_K_M.gguf -ngl 99 \
  -p "<user>\nHow do I configure Redis caching for high availability? Reason in Grug Speech (<think>...).\n<assistant>\n<think>\n" \
  -n 256

🛠️ Quickstart with Ollama

Create a Modelfile:

FROM ./MiniCPM5-2B-GrugSpeech-Q4_K_M.gguf
TEMPLATE """<user>
{{ .Prompt }} Reason in Grug Speech (<think>...).
<assistant>
<think>
"""
PARAMETER stop "<|im_end|>"
PARAMETER stop "<assistant>"

Then create and run:

ollama create minicpm5-grug -f Modelfile
ollama run minicpm5-grug

📜 Citation & Credits

Base model: OpenBMB MiniCPM5-2B
Training & Grugification by Novasaki
GitHub Project: Nov4Saki/grug-speech-reasoning

🏆 Full-Suit Master Leaderboard: 11 Models Under Strict Single-Model Isolation

Evaluated across 7 multi-lingual and multi-complexity scenarios on the live enterprise Data Analysis Agent (sales.xlsx, 100,300 rows) with 100% VRAM purge between runs:

Model Name	Architecture Family	Quant Format	Memory (GB)	Pass Rate	Avg Wall (s)	Gen Speed (t/s)	Prefill (t/s)	Avg CoT (tok)	Schema OK
`bonsai-8b-prismml`	PrismML 1-Bit	`Q1_0`	1.08 GB	6/7 (85.7%)	1.16s	419.2 t/s	8,830.6 t/s	0.0 t	7/7 (100%)
`qwen3.5-4b-grugspeech-native`	Qwen 3.5 4B	`Q4_K_M`	2.60 GB	5/7 (71.4%)	1.85s	270.2 t/s	5,808.6 t/s	38.3 t	6/7 (85.7%)
`bonsai-4b-prismml`	PrismML 1-Bit	`Q1_0`	0.53 GB	5/7 (71.4%)	1.16s	476.3 t/s	11,886.8 t/s	0.0 t	7/7 (100%)
`bonsai-1.7b-prismml`	PrismML 1-Bit	`Q1_0`	0.23 GB	5/7 (71.4%)	1.04s	691.5 t/s ⚡	16,878.5 t/s ⚡	0.0 t	7/7 (100%)
`qwen3.5-4b-base`	Qwen 3.5 4B	`Q4_K_M`	2.55 GB	3/7 (42.9%)	2.27s	269.0 t/s	5,271.4 t/s	97.3 t	3/7 (42.9%)
`gemma-4-e2b-grugspeech-native`	Google Gemma 4	`Q4_K_M`	3.20 GB	2/7 (28.6%)	2.23s	306.0 t/s	5,128.8 t/s	21.7 t	3/7 (42.9%)
`minicpm5-2b-grugspeech-native`	OpenBMB MiniCPM	`Q4_K_M`	1.49 GB	2/7 (28.6%)	1.14s	390.4 t/s	12,771.7 t/s	31.0 t	3/7 (42.9%)
`gemma-4-e2b-base`	Google Gemma 4	`Q4_K_M`	2.89 GB	1/7 (14.3%)	2.44s	321.7 t/s	3,294.8 t/s	66.9 t	1/7 (14.3%)
`qwen3.5-2b-base`	Qwen 3.5 2B	`Q4_K_M`	1.19 GB	1/7 (14.3%)	1.87s	453.8 t/s	7,775.7 t/s	79.3 t	3/7 (42.9%)
`nanbeige4.2-3b-grugspeech-native`	BOSS Zhipin 4.2	`Q4_K_M`	2.39 GB	1/7 (14.3%)	2.33s	220.1 t/s	8,487.8 t/s	46.0 t	1/7 (14.3%)
`qwen3.5-2b-grugspeech-native`	Qwen 3.5 2B	`Q4_K_M`	1.20 GB	0/7 ( 0.0%)	1.56s	450.9 t/s	8,853.0 t/s	70.6 t	1/7 (14.3%)

📈 Live End-to-End Execution Benchmark on `sales.xlsx` (100,300 Rows)

Evaluated against the live pandas dataframe of sales.xlsx (100,300 rows x 12 columns, 6.4 MB) using the application's native toolchain (aggregate_tool, filter_tool, view_tool, transform_tool, chart_tool):

Model Name	Architecture Category	Real Execution Pass Rate	Wall Latency (s)	Decode Speed (t/s)	Prompt Prefill (t/s)	Avg CoT (tok)	Avg Total Tokens
`gemma-4-e2b-grugspeech-native`	Grug Native	4/6 (66.7%) 🥇	2.30s	308.3 t/s	11,376.3 t/s	68.2 t	133.0 t
`bonsai-4b-prismml`	Base / 1-Bit	4/6 (66.7%) 🥇	1.16s ⚡	449.2 t/s	17,353.8 t/s	0.0 t	50.3 t
`bonsai-8b-prismml`	Base / 1-Bit	3/6 (50.0%)	1.23s	387.8 t/s	12,292.9 t/s	0.0 t	49.8 t
`qwen3.5-4b-grugspeech-native`	Grug Native	3/6 (50.0%)	1.84s	269.2 t/s	10,145.1 t/s	38.3 t	66.0 t
`qwen3.5-4b-base`	Base Peer	2/6 (33.3%)	2.32s	267.2 t/s	9,811.6 t/s	82.8 t	205.8 t
`gemma-4-e2b-base`	Base Peer	2/6 (33.3%)	2.46s	319.7 t/s	8,481.1 t/s	38.7 t	169.8 t
`minicpm5-2b-grugspeech-native`	Grug Native	2/6 (33.3%)	1.16s	380.0 t/s	18,933.3 t/s ⚡	39.8 t	64.2 t
`bonsai-1.7b-prismml`	Base / 1-Bit	2/6 (33.3%)	1.05s	651.8 t/s ⚡	26,843.7 t/s ⚡	0.0 t	51.7 t
`qwen3.5-2b-base`	Base Peer	1/6 (16.7%)	1.76s	458.6 t/s	15,960.9 t/s	52.0 t	224.2 t
`qwen3.5-2b-grugspeech-native`	Grug Native	0/6 (0.0%)	1.60s	458.4 t/s	17,061.8 t/s	105.7 t	152.7 t
`nanbeige4.2-3b-grugspeech-native`	Grug Native	0/6 (0.0%)	2.44s	208.5 t/s	11,261.7 t/s	0.0 t	217.0 t

Downloads last month: 89

GGUF

Model size

3B params

Architecture

llama

Hardware compatibility

4-bit

8-bit

Model tree for Novasaki/MiniCPM5-2B-GrugSpeech-Native

Base model

openbmb/MiniCPM5-2B

Adapter

(29)

this model

MiniCPM5-2B-GrugSpeech-Native

⚡ Key Highlights & Benchmark Metrics

🏎️ Cross-Family Comparison (Grug Speech Native)

📦 Quantized GGUF Binaries

💡 Prompt Format & Trigger

Sample Output

🛠️ Quickstart with llama.cpp

🛠️ Quickstart with Ollama

📜 Citation & Credits

🏆 Full-Suit Master Leaderboard: 11 Models Under Strict Single-Model Isolation

📈 Live End-to-End Execution Benchmark on sales.xlsx (100,300 Rows)

Model tree for Novasaki/MiniCPM5-2B-GrugSpeech-Native

📈 Live End-to-End Execution Benchmark on `sales.xlsx` (100,300 Rows)