Instructions to use RepublicOfKorokke/Qwen3-Coder-Next-REAP-48B-A3B-oQ3.5 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use RepublicOfKorokke/Qwen3-Coder-Next-REAP-48B-A3B-oQ3.5 with MLX:

# Make sure mlx-lm is installed
# pip install --upgrade mlx-lm

# Generate text with mlx-lm
from mlx_lm import load, generate

model, tokenizer = load("RepublicOfKorokke/Qwen3-Coder-Next-REAP-48B-A3B-oQ3.5")

prompt = "Write a story about Einstein"
messages = [{"role": "user", "content": prompt}]
prompt = tokenizer.apply_chat_template(
    messages, add_generation_prompt=True
)

text = generate(model, tokenizer, prompt=prompt, verbose=True)

Notebooks
Google Colab
Kaggle
Local Apps
LM Studio

Pi new

How to use RepublicOfKorokke/Qwen3-Coder-Next-REAP-48B-A3B-oQ3.5 with Pi:

Start the MLX server

# Install MLX LM:
uv tool install mlx-lm
# Start a local OpenAI-compatible server:
mlx_lm.server --model "RepublicOfKorokke/Qwen3-Coder-Next-REAP-48B-A3B-oQ3.5"

Configure the model in Pi

# Install Pi:
npm install -g @mariozechner/pi-coding-agent
# Add to ~/.pi/agent/models.json:
{
  "providers": {
    "mlx-lm": {
      "baseUrl": "http://localhost:8080/v1",
      "api": "openai-completions",
      "apiKey": "none",
      "models": [
        {
          "id": "RepublicOfKorokke/Qwen3-Coder-Next-REAP-48B-A3B-oQ3.5"
        }
      ]
    }
  }
}

Run Pi

# Start Pi in your project directory:
pi

Hermes Agent new

How to use RepublicOfKorokke/Qwen3-Coder-Next-REAP-48B-A3B-oQ3.5 with Hermes Agent:

Start the MLX server

# Install MLX LM:
uv tool install mlx-lm
# Start a local OpenAI-compatible server:
mlx_lm.server --model "RepublicOfKorokke/Qwen3-Coder-Next-REAP-48B-A3B-oQ3.5"

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 RepublicOfKorokke/Qwen3-Coder-Next-REAP-48B-A3B-oQ3.5

Run Hermes

hermes

MLX LM

How to use RepublicOfKorokke/Qwen3-Coder-Next-REAP-48B-A3B-oQ3.5 with MLX LM:

Generate or start a chat session

# Install MLX LM
uv tool install mlx-lm
# Interactive chat REPL
mlx_lm.chat --model "RepublicOfKorokke/Qwen3-Coder-Next-REAP-48B-A3B-oQ3.5"

Run an OpenAI-compatible server

# Install MLX LM
uv tool install mlx-lm
# Start the server
mlx_lm.server --model "RepublicOfKorokke/Qwen3-Coder-Next-REAP-48B-A3B-oQ3.5"
# Calling the OpenAI-compatible server with curl
curl -X POST "http://localhost:8000/v1/chat/completions" \
   -H "Content-Type: application/json" \
   --data '{
     "model": "RepublicOfKorokke/Qwen3-Coder-Next-REAP-48B-A3B-oQ3.5",
     "messages": [
       {"role": "user", "content": "Hello"}
     ]
   }'

Qwen3-Coder-Next-REAP-48B-A3B-oQ3.5

This model was quantized using oQ mixed-precision quantization.

Quantization details

Model type: qwen3_next
Bits: 3
Group size: 64
Format: MLX safetensors

Benchmark

Model	File size	MMLU	JMMLU	HELLASWAG	GSM8K	ARC_CHALLENGE
Qwen3-Coder-Next-REAP-48B-A3B-oQ3	20.86 GB	64.3%	56.0%	80.3%	94.3%	85.7%
Qwen3-Coder-Next-REAP-48B-A3B-oQ3.5	22.41 GB	65.0%	57.7%	83.3%	95.0%	84.3%
Qwen3-Coder-Next-REAP-48B-A3B-oQ4	26.20 GB	62.0%	57.3%	81.0%	95.3%	85.7%
Qwen3-Coder-Next-REAP-48B-A3B-mlx-mxfp4	24.20 GB	62.7%	54.7%	76.0%	93.3%	85.7%
Qwen3-Coder-Next-REAP-q4-mlx	35.18 GB	66.3%	62.0%	78.0%	94.3%	84.0%

Detail

Model	Benchmark	Accuracy	Correct	Total	Time(s)
Qwen3-Coder-Next-REAP-48B-A3B-mlx-mxfp4	HELLASWAG	76.0%	228	300	293.8
Qwen3-Coder-Next-REAP-48B-A3B-mlx-mxfp4	ARC_CHALLENGE	85.7%	257	300	180.9
Qwen3-Coder-Next-REAP-48B-A3B-mlx-mxfp4	GSM8K	93.3%	280	300	1339.4
Qwen3-Coder-Next-REAP-48B-A3B-mlx-mxfp4	HELLASWAG	76.0%	228	300	293.8
Qwen3-Coder-Next-REAP-48B-A3B-mlx-mxfp4	ARC_CHALLENGE	85.7%	257	300	180.9
Qwen3-Coder-Next-REAP-48B-A3B-mlx-mxfp4	GSM8K	93.3%	280	300	1339.4
Qwen3-Coder-Next-REAP-48B-A3B-mlx-mxfp4	MMLU	62.7%	188	300	613.2
Qwen3-Coder-Next-REAP-48B-A3B-mlx-mxfp4	JMMLU	54.7%	164	300	225.9
Qwen3-Coder-Next-REAP-48B-A3B-oQ3	MMLU	64.3%	193	300	671.1
Qwen3-Coder-Next-REAP-48B-A3B-oQ3	JMMLU	56.0%	168	300	240.4
Qwen3-Coder-Next-REAP-48B-A3B-oQ3	HELLASWAG	80.3%	241	300	288.5
Qwen3-Coder-Next-REAP-48B-A3B-oQ3	ARC_CHALLENGE	85.7%	257	300	182.7
Qwen3-Coder-Next-REAP-48B-A3B-oQ3	GSM8K	94.3%	283	300	1261.6
Qwen3-Coder-Next-REAP-48B-A3B-oQ3.5	MMLU	65.0%	195	300	711.8
Qwen3-Coder-Next-REAP-48B-A3B-oQ3.5	JMMLU	57.7%	173	300	252.7
Qwen3-Coder-Next-REAP-48B-A3B-oQ3.5	HELLASWAG	83.3%	250	300	325
Qwen3-Coder-Next-REAP-48B-A3B-oQ3.5	ARC_CHALLENGE	84.3%	253	300	179.8
Qwen3-Coder-Next-REAP-48B-A3B-oQ3.5	GSM8K	95.0%	285	300	1377.4
Qwen3-Coder-Next-REAP-48B-A3B-oQ4	MMLU	62.0%	186	300	703.4
Qwen3-Coder-Next-REAP-48B-A3B-oQ4	JMMLU	57.3%	172	300	245.7
Qwen3-Coder-Next-REAP-48B-A3B-oQ4	HELLASWAG	81.0%	243	300	307.2
Qwen3-Coder-Next-REAP-48B-A3B-oQ4	ARC_CHALLENGE	85.7%	257	300	180.4
Qwen3-Coder-Next-REAP-48B-A3B-oQ4	GSM8K	95.3%	286	300	1411.5
Qwen3-Coder-Next-REAP-q4-mlx	MMLU	66.3%	199	300	858.2
Qwen3-Coder-Next-REAP-q4-mlx	JMMLU	62.0%	186	300	278.9
Qwen3-Coder-Next-REAP-q4-mlx	HELLASWAG	78.0%	234	300	377
Qwen3-Coder-Next-REAP-q4-mlx	ARC_CHALLENGE	84.0%	252	300	181.7
Qwen3-Coder-Next-REAP-q4-mlx	GSM8K	94.3%	283	300	1516.3

Downloads last month: 777

Safetensors

Model size

7B params

Tensor type

U32

BF16

MLX

Hardware compatibility

3-bit

Model tree for RepublicOfKorokke/Qwen3-Coder-Next-REAP-48B-A3B-oQ3.5

Base model

Qwen/Qwen3-Coder-Next

Quantized

lovedheart/Qwen3-Coder-Next-REAP-48B-A3B-GGUF

Quantized

(5)

this model