Instructions to use stealth-model/space-bunny-alpha with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Notebooks
Google Colab
Kaggle
Local Apps Settings

How to use stealth-model/space-bunny-alpha 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 stealth-model/space-bunny-alpha:IQ2_XXS
# Run inference directly in the terminal:
llama cli -hf stealth-model/space-bunny-alpha:IQ2_XXS

Install from WinGet (Windows)

winget install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama serve -hf stealth-model/space-bunny-alpha:IQ2_XXS
# Run inference directly in the terminal:
llama cli -hf stealth-model/space-bunny-alpha:IQ2_XXS

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 stealth-model/space-bunny-alpha:IQ2_XXS
# Run inference directly in the terminal:
./build/bin/llama cli -hf stealth-model/space-bunny-alpha:IQ2_XXS

Use Docker

docker model run hf.co/stealth-model/space-bunny-alpha:IQ2_XXS

Llama new
LM Studio
Jan

vLLM

How to use stealth-model/space-bunny-alpha with vLLM:

Install from pip and serve model

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

Use Docker

docker model run hf.co/stealth-model/space-bunny-alpha:IQ2_XXS

Ollama
How to use stealth-model/space-bunny-alpha with Ollama:
```
ollama run hf.co/stealth-model/space-bunny-alpha:IQ2_XXS
```
Unsloth Desktop

How to use stealth-model/space-bunny-alpha with Pi:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf stealth-model/space-bunny-alpha:IQ2_XXS

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": "stealth-model/space-bunny-alpha:IQ2_XXS"
        }
      ]
    }
  }
}

Run Pi

# Start Pi in your project directory:
pi

Docker Model Runner
How to use stealth-model/space-bunny-alpha with Docker Model Runner:
```
docker model run hf.co/stealth-model/space-bunny-alpha:IQ2_XXS
```

Lemonade

How to use stealth-model/space-bunny-alpha with Lemonade:

Pull the model

# Download Lemonade from https://lemonade-server.ai/
lemonade pull stealth-model/space-bunny-alpha:IQ2_XXS

Run and chat with the model

lemonade run user.space-bunny-alpha-IQ2_XXS

List all available models

lemonade list

Hermes Agent

How to use stealth-model/space-bunny-alpha with Hermes Agent:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf stealth-model/space-bunny-alpha:IQ2_XXS

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 stealth-model/space-bunny-alpha:IQ2_XXS

Run Hermes

hermes

Atomic Chat

OpenClaw

How to use stealth-model/space-bunny-alpha with OpenClaw:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf stealth-model/space-bunny-alpha:IQ2_XXS

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 "stealth-model/space-bunny-alpha:IQ2_XXS" \
  --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"

Space Bunny Alpha. Revealed: an early MiniMax M3.1 Flash. The stealth model MiniMax just unmasked.

Our token counts matched MiniMax on all 24 test phrases, 20 days before MiniMax named Space Bunny Alpha an early M3.1 Flash.

By Stealth Models.

StealthMark 1.8.10 results for Space Bunny Alpha: benchmark score 45.5, General Intelligence 35.7, Visual Intelligence 64.4. Strongest at turning a brief into an original character: 78.9 on Selfie.

Every StealthMark result of Space Bunny Alpha: Selfie 78.9, Raccoon 74.3, Beach 67.8, Mona Lisa 49.9, Ground Truth 42.7, Foresight 35.8, Latch 34.4, Newton's Mirror 24.9.

See how the early version compares with the final release

The case file. First seen 23 Sep 2026, revealed 19 Oct 2026, developer MiniMax, 1M-token context, text, image and video input, tool calls, about 300 tokens per second in our tests. 23 Sep 2026: appears with no developer's name. 29 Sep 2026, our finding: our token-count test matches MiniMax M3 on all 24 test phrases. 19 Oct 2026: MiniMax names it an early version of MiniMax M3.1 Flash.

See the token-count test

Facts as text

Fact	Value
First seen	23 Sep 2026
Revealed	19 Oct 2026
Developer	MiniMax
Context	1M tokens
Input	Text, image, video
Tool calls	Yes
Speed in our tests	about 300 tok/s
Access	Stealth access ended; MiniMax M3.1 Flash is available

Run it

The final release, MiniMax-M3.1-Flash, is open for non-commercial use under the MiniMax Community License (LICENSE). Its GGUF files are here.

File	Size	Best for
`IQ2_XXS/`	135 GB in 4 parts	The smallest size we recommend: 192 GB machines, or a 32 GB GPU with 128 GB of RAM
`dspark-MiniMax-M3.1-Flash-Q8_0.gguf`	2.1 GB	DSpark draft: about 1.5x faster on code and math
`mmproj-F16.gguf`	1.7 GB	Image and video input, full precision
`mmproj-Q8_0.gguf`	0.93 GB	Image and video input, near lossless
`mmproj-Q6_K.gguf`	0.72 GB	Image and video input, smaller
`mmproj-Q4_K.gguf`	0.49 GB	Image and video input, smallest

hf download stealth-model/space-bunny-alpha --local-dir minimax-m3.1-flash \
  --include "IQ2_XXS/*" "mmproj-Q8_0.gguf" "dspark-MiniMax-M3.1-Flash-Q8_0.gguf"

llama-server \
  -m minimax-m3.1-flash/IQ2_XXS/MiniMax-M3.1-Flash-IQ2_XXS-00001-of-00004.gguf \
  --mmproj minimax-m3.1-flash/mmproj-Q8_0.gguf \
  -md minimax-m3.1-flash/dspark-MiniMax-M3.1-Flash-Q8_0.gguf \
  --spec-type draft-dspark --spec-draft-n-max 1 --spec-draft-p-min 0 -np 1 \
  -ngl all -ngld all --n-cpu-moe 60 -fa on --jinja --temp 1.0 --top-p 0.95

--n-cpu-moe 60 keeps all 60 expert layers in system RAM. Lower it until your GPU is full. On a 192 GB Mac, leave it out.

Model guide · StealthMark leaderboard
Text and images: CC BY 4.0, stealthmodels.com.

Downloads last month: 633

GGUF

Model size

426B params

Architecture

minimax-m3

Hardware compatibility

2-bit

8-bit

Model tree for stealth-model/space-bunny-alpha

Base model

MiniMaxAI/MiniMax-M3.1-Flash

Quantized

(39)

this model

stealth-model
/

space-bunny-alpha

Run it

Model tree for stealth-model/space-bunny-alpha

Space using stealth-model/space-bunny-alpha 1