Instructions to use ruv/ruvltra-claude-code with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use ruv/ruvltra-claude-code with MambaSSM:

from mamba_ssm import MambaLMHeadModel

model = MambaLMHeadModel.from_pretrained("ruv/ruvltra-claude-code")

llama-cpp-python

How to use ruv/ruvltra-claude-code with llama-cpp-python:

# !pip install llama-cpp-python

from llama_cpp import Llama

llm = Llama.from_pretrained(
	repo_id="ruv/ruvltra-claude-code",
	filename="ruvltra-claude-code-0.5b-q4_k_m.gguf",
)

llm.create_chat_completion(
	messages = [
		{
			"role": "user",
			"content": "What is the capital of France?"
		}
	]
)

Notebooks
Google Colab
Kaggle
Local Apps

llama.cpp

How to use ruv/ruvltra-claude-code with llama.cpp:

Install from brew

brew install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama-server -hf ruv/ruvltra-claude-code:Q4_K_M
# Run inference directly in the terminal:
llama-cli -hf ruv/ruvltra-claude-code:Q4_K_M

Install from WinGet (Windows)

winget install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama-server -hf ruv/ruvltra-claude-code:Q4_K_M
# Run inference directly in the terminal:
llama-cli -hf ruv/ruvltra-claude-code: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 ruv/ruvltra-claude-code:Q4_K_M
# Run inference directly in the terminal:
./llama-cli -hf ruv/ruvltra-claude-code: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 ruv/ruvltra-claude-code:Q4_K_M
# Run inference directly in the terminal:
./build/bin/llama-cli -hf ruv/ruvltra-claude-code:Q4_K_M

Use Docker

docker model run hf.co/ruv/ruvltra-claude-code:Q4_K_M

LM Studio
Jan

vLLM

How to use ruv/ruvltra-claude-code with vLLM:

Install from pip and serve model

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

Use Docker

docker model run hf.co/ruv/ruvltra-claude-code:Q4_K_M

Ollama
How to use ruv/ruvltra-claude-code with Ollama:
```
ollama run hf.co/ruv/ruvltra-claude-code:Q4_K_M
```

Unsloth Studio new

How to use ruv/ruvltra-claude-code with Unsloth Studio:

Install Unsloth Studio (macOS, Linux, WSL)

curl -fsSL https://unsloth.ai/install.sh | sh
# Run unsloth studio
unsloth studio -H 0.0.0.0 -p 8888
# Then open http://localhost:8888 in your browser
# Search for ruv/ruvltra-claude-code to start chatting

Install Unsloth Studio (Windows)

irm https://unsloth.ai/install.ps1 | iex
# Run unsloth studio
unsloth studio -H 0.0.0.0 -p 8888
# Then open http://localhost:8888 in your browser
# Search for ruv/ruvltra-claude-code to start chatting

Using HuggingFace Spaces for Unsloth

# No setup required
# Open https://huggingface.co/spaces/unsloth/studio in your browser
# Search for ruv/ruvltra-claude-code to start chatting

Docker Model Runner
How to use ruv/ruvltra-claude-code with Docker Model Runner:
```
docker model run hf.co/ruv/ruvltra-claude-code:Q4_K_M
```

Lemonade

How to use ruv/ruvltra-claude-code with Lemonade:

Pull the model

# Download Lemonade from https://lemonade-server.ai/
lemonade pull ruv/ruvltra-claude-code:Q4_K_M

Run and chat with the model

lemonade run user.ruvltra-claude-code-Q4_K_M

List all available models

lemonade list

Interesting model

by FlameF0X - opened Feb 4

Discussion

FlameF0X

Feb 4

•

edited Feb 4

So... i try it and it feels promising. It's fast on my laptop with a intel i3-6006u CPU and 12GB of RAM.

You, I, or someone, should run a benchmark on this model.

I did test it on code. It's interesting, it hallucinates at complex tasks-- but its good at simple assisting.

FlameF0X

Feb 6

i ran a benchmark

CompactAI

Mar 8

kinda sad

Ayedunno

Apr 1

I'm getting a ton of hallucination

CompactAI

Apr 1

Well, it is a crazy small model

FlameF0X

Apr 1

I'm getting a ton of hallucination

Really?

CompactAI

Apr 1

Yes. Its a 0.5 Billion parameter model. 500M. That's crazy small.

FlameF0X

Apr 1

I know that its small. Qwen2-0.5B-Instruct behaves the same.

But not all SLM hallucinate the same. Like LFM2.5 350M, its smaller than qwen3, but it hallucinates less when using it than the previous version. (This sounds kinda out of context 😐 )

CompactAI

Apr 1

•

edited Apr 1

Guess the maker of this didn't do RL for hallucinations

FlameF0X

Apr 1

•

edited Apr 1

do*

But yes. I think the RL for preventing hallucinations wasn't in the model road map.

Also, hallucination can be solved not only with RL, but also with a proper training. If the model is under-trained, it tends to hallucinate more often. (Saying from my own experience)

FlameF0X

Apr 1

I do suspect that this is just Qwen3 0.5B but with a different name.

CompactAI

Apr 1

•

edited Apr 1

It would be nice if they provided their own inference script so we could see it actually evolve, because llama.cpp (or vllm) don't actually have this feature

Edit: Oops, guess it does

FlameF0X

Apr 1

this?

use ruvllm::sona::SonaConfig;

let config = SonaConfig {
    micro_lora_rank: 2,
    base_lora_rank: 8,
    learning_rate: 0.001,
    ewc_lambda: 0.5,  // Memory protection strength
    pattern_threshold: 0.75,
    ..Default::default()
};

Upload images, audio, and videos by dragging in the text input, pasting, or clicking here.

Tap or paste here to upload images

· Sign up or log in to comment