Instructions to use ARO-Lang/aro-coder-6bit with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use ARO-Lang/aro-coder-6bit 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("ARO-Lang/aro-coder-6bit")

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 Settings
LM Studio

How to use ARO-Lang/aro-coder-6bit with Pi:

Start the MLX server

# Install MLX LM:
uv tool install mlx-lm
# Start a local OpenAI-compatible server:
mlx_lm.server --model "ARO-Lang/aro-coder-6bit"

Configure the model in Pi

# Install Pi:
npm install -g @earendil-works/pi-coding-agent
# Add to ~/.pi/agent/models.json:
{
  "providers": {
    "mlx-lm": {
      "baseUrl": "http://localhost:8080/v1",
      "api": "openai-completions",
      "apiKey": "none",
      "models": [
        {
          "id": "ARO-Lang/aro-coder-6bit"
        }
      ]
    }
  }
}

Run Pi

# Start Pi in your project directory:
pi

MLX LM

How to use ARO-Lang/aro-coder-6bit with MLX LM:

Generate or start a chat session

# Install MLX LM
uv tool install mlx-lm
# Interactive chat REPL
mlx_lm.chat --model "ARO-Lang/aro-coder-6bit"

Run an OpenAI-compatible server

# Install MLX LM
uv tool install mlx-lm
# Start the server
mlx_lm.server --model "ARO-Lang/aro-coder-6bit"
# Calling the OpenAI-compatible server with curl
curl -X POST "http://localhost:8000/v1/chat/completions" \
   -H "Content-Type: application/json" \
   --data '{
     "model": "ARO-Lang/aro-coder-6bit",
     "messages": [
       {"role": "user", "content": "Hello"}
     ]
   }'

Hermes Agent

How to use ARO-Lang/aro-coder-6bit 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 "ARO-Lang/aro-coder-6bit"

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 ARO-Lang/aro-coder-6bit

Run Hermes

hermes

Atomic Chat

OpenClaw

How to use ARO-Lang/aro-coder-6bit with OpenClaw:

Start the MLX server

# Install MLX LM:
uv tool install mlx-lm
# Start a local OpenAI-compatible server:
mlx_lm.server --model "ARO-Lang/aro-coder-6bit"

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 "ARO-Lang/aro-coder-6bit" \
  --custom-provider-id mlx-lm \
  --custom-compatibility openai \
  --custom-text-input \
  --accept-risk \
  --skip-health

Run OpenClaw

openclaw agent --local --agent main --message "Hello from Hugging Face"

KrisSimon commited on Jul 28

Commit

1eb5b5a

verified ·

1 Parent(s): 94d3e39

Release v1.1.0 — ARO Coder 6-bit (distill_student, 12316118709fd2ce)

Browse files

Files changed (1) hide show

README.md +4 -4

README.md CHANGED Viewed

@@ -6,7 +6,7 @@ tags:
   - code-generation
   - dsl
   - mlx
-  - 4-bit
   - lora
   - fine-tuned
 base_model: mlx-community/Qwen3-Coder-30B-A3B-Instruct-4bit
@@ -27,7 +27,7 @@ ARO is a domain-specific language where every statement follows the pattern:
 | **Checksum** | `12316118709fd2ce` |
 | **Base model** | [mlx-community/Qwen3-Coder-30B-A3B-Instruct-4bit](https://huggingface.co/mlx-community/Qwen3-Coder-30B-A3B-Instruct-4bit) |
 | **Teacher source** | distill_student (30B MoE teacher distilled to 8B student) |
-| **Quantization** | 4-bit MLX, group size 64 |
 | **Language** | ARO |
 | **Training samples** | 7760 |
@@ -53,7 +53,7 @@ ARO is a domain-specific language where every statement follows the pattern:
 - **Happy-path DSL only** — ARO code deliberately contains no error handling;
   do not expect defensive code from this model.
-- **4-bit quantization** — small quality loss vs the fused model is expected;
   the promotion gate bounds the degradation (see the table above when both
   columns are present).
 - **Verb hallucination at high temperatures** — keep temperature ≤ 0.3 for
@@ -144,7 +144,7 @@ This model was trained with the ARO training pipeline:
 | Version | Date | Source | Checksum |
 |---|---|---|---|
-| v1.1.0 | 2026-07-27 | distill_student | `12316118709fd2ce` |
 Every release is tagged on the Hub — load an older version with
 `load("ARO-Lang/aro-coder-6bit", revision="v<version>")` or report issues against

   - code-generation
   - dsl
   - mlx
+  - 6-bit
   - lora
   - fine-tuned
 base_model: mlx-community/Qwen3-Coder-30B-A3B-Instruct-4bit
 | **Checksum** | `12316118709fd2ce` |
 | **Base model** | [mlx-community/Qwen3-Coder-30B-A3B-Instruct-4bit](https://huggingface.co/mlx-community/Qwen3-Coder-30B-A3B-Instruct-4bit) |
 | **Teacher source** | distill_student (30B MoE teacher distilled to 8B student) |
+| **Quantization** | 6-bit MLX, group size 32 |
 | **Language** | ARO |
 | **Training samples** | 7760 |
 - **Happy-path DSL only** — ARO code deliberately contains no error handling;
   do not expect defensive code from this model.
+- **6-bit quantization** — small quality loss vs the fused model is expected;
   the promotion gate bounds the degradation (see the table above when both
   columns are present).
 - **Verb hallucination at high temperatures** — keep temperature ≤ 0.3 for
 | Version | Date | Source | Checksum |
 |---|---|---|---|
+| v1.1.0 | 2026-07-28 | distill_student | `12316118709fd2ce` |
 Every release is tagged on the Hub — load an older version with
 `load("ARO-Lang/aro-coder-6bit", revision="v<version>")` or report issues against