Instructions to use Yo-Purple/Meditor-3-4B-Universal with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Notebooks
Google Colab
Kaggle
Local Apps Settings

How to use Yo-Purple/Meditor-3-4B-Universal 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 Yo-Purple/Meditor-3-4B-Universal:Q4_K_M
# Run inference directly in the terminal:
llama cli -hf Yo-Purple/Meditor-3-4B-Universal:Q4_K_M

Install from WinGet (Windows)

winget install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama serve -hf Yo-Purple/Meditor-3-4B-Universal:Q4_K_M
# Run inference directly in the terminal:
llama cli -hf Yo-Purple/Meditor-3-4B-Universal: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 Yo-Purple/Meditor-3-4B-Universal:Q4_K_M
# Run inference directly in the terminal:
./build/bin/llama cli -hf Yo-Purple/Meditor-3-4B-Universal:Q4_K_M

Use Docker

docker model run hf.co/Yo-Purple/Meditor-3-4B-Universal:Q4_K_M

Llama new
LM Studio
Jan

vLLM

How to use Yo-Purple/Meditor-3-4B-Universal with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "Yo-Purple/Meditor-3-4B-Universal"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "Yo-Purple/Meditor-3-4B-Universal",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/Yo-Purple/Meditor-3-4B-Universal:Q4_K_M

Ollama
How to use Yo-Purple/Meditor-3-4B-Universal with Ollama:
```
ollama run hf.co/Yo-Purple/Meditor-3-4B-Universal:Q4_K_M
```
Unsloth Desktop

How to use Yo-Purple/Meditor-3-4B-Universal with Pi:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf Yo-Purple/Meditor-3-4B-Universal: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": "Yo-Purple/Meditor-3-4B-Universal:Q4_K_M"
        }
      ]
    }
  }
}

Run Pi

# Start Pi in your project directory:
pi

Docker Model Runner
How to use Yo-Purple/Meditor-3-4B-Universal with Docker Model Runner:
```
docker model run hf.co/Yo-Purple/Meditor-3-4B-Universal:Q4_K_M
```

Lemonade

How to use Yo-Purple/Meditor-3-4B-Universal with Lemonade:

Pull the model

# Download Lemonade from https://lemonade-server.ai/
lemonade pull Yo-Purple/Meditor-3-4B-Universal:Q4_K_M

Run and chat with the model

lemonade run user.Meditor-3-4B-Universal-Q4_K_M

List all available models

lemonade list

Hermes Agent

How to use Yo-Purple/Meditor-3-4B-Universal with Hermes Agent:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf Yo-Purple/Meditor-3-4B-Universal: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 Yo-Purple/Meditor-3-4B-Universal:Q4_K_M

Run Hermes

hermes

Atomic Chat

OpenClaw

How to use Yo-Purple/Meditor-3-4B-Universal with OpenClaw:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf Yo-Purple/Meditor-3-4B-Universal: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 "Yo-Purple/Meditor-3-4B-Universal: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"

Meditor-3-4B-Universal — Standalone `k2-horizon` MoE GGUF Edition

Meditor-3-4B-Universal compiles the Yo-Purple/Meditor-3-4B-MoE system into zero-dependency k2-horizon Mixture-of-Experts GGUF checkpoints runnable out-of-the-box in llama.cpp, LM Studio, KoboldCpp, and Ollama.

Each transformer layer preserves the intact Qwen3-4B dense FFN as a shared expert (d_ff=9728) alongside 8 compact routed specialist FFN experts (d_ff=64, top-2 normalized softmax routing per token), distilled from the native LoRA expert weights and PromptRouter.

Note on Benchmarks: The 7.31 / 10 (69.0% pass, Rank #1 / 36) UniBench-100 score was measured on the full native dynamic prompt-routed Meditor-3-4B-MoE model. This repository provides the compiled standalone k2-horizon GGUF approximation (top-2 per-token FFN experts) for universal local runtime portability.

Meditor Model Series

Generation & Variant	Hugging Face Repo	Base Backbone	Params (Total / Active)	UniBench-100 Mean (/10)	Pass Rate	VRAM Footprint
Meditor 1 (Native MoE)	`Yo-Purple/Meditor-1-2.2B-A500M-MoE`	Qwen2.5-0.5B	2.22B / 0.57B	2.09 (37/100) · 2.42 (Best M1)	1.0% (Best M1)	4.16 GB (BF16)
Meditor 1 (Universal)	`Yo-Purple/Meditor-1-500M-Universal`	Qwen2.5-0.5B (`qwen2`)	0.49B / 0.49B	2.42 (100/100 BF16)	1.0%	0.93 GB / GGUF
Meditor 2 (Native HMoE)	`Yo-Purple/Meditor-2-3.4B-A1.8B-HMoE`	SmolLM2-1.7B	3.41B / 1.83B	3.28 (37/100) · 4.12 (Best M2)	22.0% (Best M2)	6.42 GB (BF16)
Meditor 2 (Universal)	`Yo-Purple/Meditor-2-3.4B-Universal`	SmolLM2-1.7B (`qwen2_moe`)	3.41B / 1.83B	4.12 (100/100 BF16) · 4.00 (Q4_K_M)	22.0% / 18.0%	6.38 GB / GGUF
Meditor 3 (Native MoE)	`Yo-Purple/Meditor-3-4B-MoE`	Qwen3-4B (`meditor3`)	4.16B / 4.04B–4.07B	7.31 (100/100, Rank #1/36)	69.0% (7.0% Uniq)	5.43 GiB peak (`bf16_low_vram`)
Meditor 3 (Universal GGUF) (This Model)	`Yo-Purple/Meditor-3-4B-Universal`	Qwen3-4B (`k2-horizon`)	4.16B (`BF16`) · `Q8_0` · `Q4_K_M`	Compiled GGUF from M3 (`7.31` native)	LM Studio / llama.cpp	4.37–5.95 GiB (quantized)

Universal LLM Benchmark (`UniBench-100`) Cross-Generation Comparison

All benchmark comparisons below come directly from the unified UniBench-100 evaluation suite (100 open-ended synthesis questions across 10 cognitive domains, graded out of 10.0 with pass threshold >= 6.0 by gemini-3.8-flash).

1. Cross-Generation Progression (Full 100/100 Runs)

Rank (of 36 Runs)	Model Run	Series Generation	Evaluated	Mean Score (/10)	Pass Rate (>=6.0)	Unique Solve Rate	Eval Speed
#1	Meditor-3-4B-MoE (Released)	Meditor 3 (Native)	100 / 100	7.31	69.0%	7.0%	35.7 tok/s
#2	Meditor-3 Prior Uncontested (BF16)	Meditor 3 (Prior R&D)	100 / 100	4.62	30.0%	1.0%	35.7 tok/s
#5	Meditor 3 Beta (BF16)	Meditor 3 (Prior R&D)	100 / 100	4.31	22.0%	1.0%	19.0 tok/s
#11	Meditor 2 Universal (BF16)	Meditor 2 Best	100 / 100	4.12	22.0%	0.0%	27.5 tok/s
#18	Meditor 2 Universal (Q4_K_M)	Meditor 2 Quantized	100 / 100	4.00	18.0%	0.0%	141.2 tok/s
#29	Meditor 2 (Native HMoE)	Meditor 2 Native	37 / 100	3.28	8.1%	0.0%	48.7 tok/s
#30	Meditor 1 Universal (BF16)	Meditor 1 Best	100 / 100	2.42	1.0%	0.0%	163.4 tok/s
#31	Meditor 1 Universal (Quant)	Meditor 1 Quantized	100 / 100	2.35	0.0%	0.0%	171.3 tok/s
#33	Meditor 1 (Native MoE)	Meditor 1 Native	37 / 100	2.09	2.7%	0.0%	141.9 tok/s

Key Takeaway: Meditor-3-4B-MoE delivers +3.19 points (+77.4% relative score jump) over the best Meditor 2 run (4.12), +4.89 points (3.02x score) over the best Meditor 1 run (2.42), and +2.69 points over the strongest earlier Meditor 3 research run (4.62).

2. Per-Domain Score Breakdown (Best Runs of Meditor 1, Meditor 2, and Meditor 3)

UniBench-100 Domain	Meditor 1 Best (`2.42`)	Meditor 2 Best (`4.12`)	Meditor 3 Released (`7.31`)	Gain vs M2	M3 Pass Rate
Critical Thinking & Argument Evaluation	3.51	5.98	9.09	`+3.11`	100.0%
Creative & Stylistic Expression	2.91	4.95	8.38	`+3.43`	90.0%
Humanities, Philosophy & Ethics	2.30	4.91	8.35	`+3.44`	80.0%
Physical & Life Sciences	1.68	3.38	7.41	`+4.03`	80.0%
Mathematics & Quantitative Estimation	2.36	1.99	7.27	`+5.28`	60.0%
Factuality & Hallucination Resistance	1.32	5.56	7.09	`+1.53`	60.0%
Real-World Tactical Strategy & Decision-Making	2.10	3.72	6.98	`+3.26`	60.0%
Instruction Following & Strict Constraints	3.65	4.62	6.82	`+2.20`	50.0%
Coding & Software Engineering	2.15	3.26	6.46	`+3.20`	70.0%
Logic & Abstract Reasoning	2.19	2.82	5.27	`+2.45`	40.0%

3. Full Leaderboard Landscape

Architecture Specifications (`Meditor-3-4B-Universal`)

Architectural Parameter	Value
GGUF Architecture	`k2-horizon` (`general.basename: Meditor-3-4B-Universal`)
Total Parameters	`4,164,938,240` (`4.16B`)
Shared Backbone Attention & Dense FFN	`36` layers, `d_model=2560`, `32Q / 8KV` heads (`d_head=128`), shared FFN `d_ff=9728`
Routed Specialist Experts	`8` routed experts per layer (`d_ff=64`), `top-2` softmax gating (`expert_weights_norm=True`)
Context Window & RoPE	`40,960` tokens (`rope.freq_base=1,000,000`)
Recommended Sampling	`temp=0.6`, `top_p=0.95`, `top_k=20` (ChatML / Qwen3 template with embedded identity fallback)

Available GGUF Files & Verified RTX 4060 (8 GB) Performance

File	Quantization	File Size	Peak CUDA VRAM	Generation Speed	Notes
`meditor-3-4b-universal-bf16.gguf`	BF16	`7.7654 GiB`	`< 8.0 GiB` weights	Full precision	Reference unquantized GGUF
`meditor-3-4b-universal-q8_0.gguf`	Q8_0	`4.1299 GiB`	`5.954 GiB`	`49.5 tok/s`	Recommended high-fidelity quant
`meditor-3-4b-universal-q4_k_m.gguf`	Q4_K_M	`2.5376 GiB`	`4.365 GiB`	`72.5 tok/s`	Recommended fast low-VRAM quant

Measured with llama-server.exe (CUDA 13, -ngl 99 -c 2048) on NVIDIA RTX 4060 8 GB, verifying default identity, system prompt override, arithmetic, code synthesis, and strict JSON output.

Quickstart (`llama.cpp` / LM Studio)

llama-server -m meditor-3-4b-universal-q8_0.gguf -ngl 99 -c 8192 --port 8080
# Or fast 4-bit:
llama-server -m meditor-3-4b-universal-q4_k_m.gguf -ngl 99 -c 8192 --port 8080

LM Studio: Search Yo-Purple/Meditor-3-4B-Universal, download Q8_0 (4.13 GiB) or Q4_K_M (2.54 GiB), and load directly with full GPU offload.
Custom System Prompts: The embedded Jinja chat_template automatically supplies a clean default Meditor-3-4B-Universal identity when no system message is present, while honoring any user-specified system prompt verbatim.

License

Released under Apache 2.0. Built on Qwen/Qwen3-4B and Yo-Purple/Meditor-3-4B-MoE.

Downloads last month: 152

GGUF

Model size

4B params

Architecture

k2-horizon

Hardware compatibility

4-bit

8-bit

16-bit

Model tree for Yo-Purple/Meditor-3-4B-Universal

Base model

Qwen/Qwen3-4B-Base

Finetuned

Qwen/Qwen3-4B

Quantized

(386)

this model

Evaluation results

Native Source Mean Score (/10) on UniBench-100 (Native Meditor-3-4B-MoE)
self-reported

7.310
Native Source Pass Rate (%) on UniBench-100 (Native Meditor-3-4B-MoE)
self-reported

69.000

Meditor-3-4B-Universal — Standalone k2-horizon MoE GGUF Edition

Meditor Model Series

Universal LLM Benchmark (UniBench-100) Cross-Generation Comparison

1. Cross-Generation Progression (Full 100/100 Runs)

2. Per-Domain Score Breakdown (Best Runs of Meditor 1, Meditor 2, and Meditor 3)

3. Full Leaderboard Landscape

Architecture Specifications (Meditor-3-4B-Universal)

Available GGUF Files & Verified RTX 4060 (8 GB) Performance

Quickstart (llama.cpp / LM Studio)

License

Model tree for Yo-Purple/Meditor-3-4B-Universal

Evaluation results

Meditor-3-4B-Universal — Standalone `k2-horizon` MoE GGUF Edition

Universal LLM Benchmark (`UniBench-100`) Cross-Generation Comparison

Architecture Specifications (`Meditor-3-4B-Universal`)

Quickstart (`llama.cpp` / LM Studio)