Image-Text-to-Text
MLX
Safetensors
modilify_mk1
diffusion
multimodal
mixture-of-experts
conversational
Instructions to use modilify/Modilify-Mk1-MLX with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- MLX
How to use modilify/Modilify-Mk1-MLX with MLX:
# Make sure mlx-vlm is installed # pip install --upgrade mlx-vlm from mlx_vlm import load, generate from mlx_vlm.prompt_utils import apply_chat_template from mlx_vlm.utils import load_config # Load the model model, processor = load("modilify/Modilify-Mk1-MLX") config = load_config("modilify/Modilify-Mk1-MLX") # Prepare input image = ["http://images.cocodataset.org/val2017/000000039769.jpg"] prompt = "Describe this image." # Apply chat template formatted_prompt = apply_chat_template( processor, config, prompt, num_images=1 ) # Generate output output = generate(model, processor, formatted_prompt, image) print(output) - Notebooks
- Google Colab
- Kaggle
- Local Apps Settings
- LM Studio
- Pi
How to use modilify/Modilify-Mk1-MLX with Pi:
Start the MLX server
# Install MLX LM: uv tool install mlx-lm # Start a local OpenAI-compatible server: mlx_lm.server --model "modilify/Modilify-Mk1-MLX"
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": "modilify/Modilify-Mk1-MLX" } ] } } }Run Pi
# Start Pi in your project directory: pi
- Hermes Agent
How to use modilify/Modilify-Mk1-MLX 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 "modilify/Modilify-Mk1-MLX"
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 modilify/Modilify-Mk1-MLX
Run Hermes
hermes
- Atomic Chat
- OpenClaw
How to use modilify/Modilify-Mk1-MLX with OpenClaw:
Start the MLX server
# Install MLX LM: uv tool install mlx-lm # Start a local OpenAI-compatible server: mlx_lm.server --model "modilify/Modilify-Mk1-MLX"
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 "modilify/Modilify-Mk1-MLX" \ --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"
| # Copyright 2026 Modilify | |
| # SPDX-License-Identifier: LicenseRef-Modilify-Open-Model-1.0 | |
| """Optional per-denoise phase timers. Enabled only with --profile.""" | |
| from __future__ import annotations | |
| from collections import defaultdict | |
| import time | |
| import mlx.core as mx | |
| PHASES = ( | |
| "latent", | |
| "attn", | |
| "moe", | |
| "lm_head", | |
| "softmax", | |
| "commit", | |
| "sync", | |
| "update_cache", | |
| ) | |
| class DenoiseProfiler: | |
| def __init__(self) -> None: | |
| self.totals = defaultdict(float) | |
| self.steps = 0 | |
| def add(self, phase: str, seconds: float) -> None: | |
| self.totals[phase] += float(seconds) | |
| def finish_step(self) -> None: | |
| self.steps += 1 | |
| def measure(self, phase: str, *arrays: mx.array): | |
| mx.eval(*arrays) | |
| started = time.perf_counter() | |
| class _Span: | |
| def __init__(self, profiler: DenoiseProfiler, name: str) -> None: | |
| self.profiler = profiler | |
| self.name = name | |
| self.started = started | |
| def done(self, *outputs: mx.array) -> None: | |
| if outputs: | |
| mx.eval(*outputs) | |
| self.profiler.add(self.name, time.perf_counter() - self.started) | |
| return _Span(self, phase) | |
| def summary(self) -> str: | |
| counted = sum(self.totals[name] for name in PHASES) | |
| lines = [ | |
| f"[profile] steps={self.steps} accounted={counted:.3f}s", | |
| ] | |
| for name in PHASES: | |
| value = self.totals[name] | |
| share = (100.0 * value / counted) if counted else 0.0 | |
| per = (value / self.steps) if self.steps else 0.0 | |
| lines.append( | |
| f"[profile] {name:12s} {value:7.3f}s {share:5.1f}% {per*1000:6.1f} ms/step" | |
| ) | |
| return "\n".join(lines) | |