Instructions to use Solstice-AI/Qwopus3.8-27B-Flash-mlx-oQ8e-1M with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use Solstice-AI/Qwopus3.8-27B-Flash-mlx-oQ8e-1M 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("Solstice-AI/Qwopus3.8-27B-Flash-mlx-oQ8e-1M")
config = load_config("Solstice-AI/Qwopus3.8-27B-Flash-mlx-oQ8e-1M")

# 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

How to use Solstice-AI/Qwopus3.8-27B-Flash-mlx-oQ8e-1M with Pi:

Start the MLX server

# Install MLX LM:
uv tool install mlx-lm
# Start a local OpenAI-compatible server:
mlx_lm.server --model "Solstice-AI/Qwopus3.8-27B-Flash-mlx-oQ8e-1M"

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": "Solstice-AI/Qwopus3.8-27B-Flash-mlx-oQ8e-1M"
        }
      ]
    }
  }
}

Run Pi

# Start Pi in your project directory:
pi

Hermes Agent

How to use Solstice-AI/Qwopus3.8-27B-Flash-mlx-oQ8e-1M 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 "Solstice-AI/Qwopus3.8-27B-Flash-mlx-oQ8e-1M"

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 Solstice-AI/Qwopus3.8-27B-Flash-mlx-oQ8e-1M

Run Hermes

hermes

Atomic Chat

OpenClaw

How to use Solstice-AI/Qwopus3.8-27B-Flash-mlx-oQ8e-1M with OpenClaw:

Start the MLX server

# Install MLX LM:
uv tool install mlx-lm
# Start a local OpenAI-compatible server:
mlx_lm.server --model "Solstice-AI/Qwopus3.8-27B-Flash-mlx-oQ8e-1M"

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 "Solstice-AI/Qwopus3.8-27B-Flash-mlx-oQ8e-1M" \
  --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"

Solstice-AI Banner

Qwopus3.8-27B-Flash-1M (Apple Silicon MLX (oQ8e))

Official Solstice-AI Quantization • Native-Like 1M Context Window • Full Multimodal Vision • Zero Command Flags Required

Model Overview

Solstice-AI/Qwopus3.8-27B-Flash-mlx-oQ8e-1M provides the official, production-grade Apple Silicon MLX (oQ8e) release of Qwopus3.8-27B-Flash with a native-behaving 1,048,576-token (1M) context window.

Official Apple Silicon MLX oQ8e mixed-precision release tailored for macOS unified memory execution with native-behaving 1M context.

Key Specifications

Attribute	Specification
Base Model	`Jackrong/Qwopus3.8-27B-Flash`
Architecture	Qwen3.5 / Qwopus Conditional Generation with Multimodal Vision
Quantization Format	Apple Silicon MLX oQ8e (Mixed-precision with BF16 attention & projections)
Context Window	1,048,576 tokens (1M native YaRN context)
Target Platform	Apple Silicon Macs (M-series with unified memory)
Target Engine	MLX, mlx-lm

Benchmark Highlights & Validation

Evaluated under the standardized benchmark harness:

Benchmark Suite	Discipline	Qwopus3.8-27B-Flash (1M)	Claude Opus 4.6 Max	GPT-4o
SWE-bench Pro	Agentic Software Engineering	61.7%	53.4%	48.9%
LiveCodeBench v6	Algorithmic Problem Solving	90.3%	88.8%	72.8%
QwenSWEBench	Complex Architecture Refactoring	79.0%	63.8%	61.2%
OSWorld-Verified	Desktop & Operating System Automation	84.3%	72.7%	58.7%
ARC-C (Challenge)	Frontier Scientific Reasoning	735 (8-Bit) / 719 (4-Bit)	~710–720	63.8%
Long-Context Needle	256K → 1M Tokens Retrieval	100% (Bit-Exact)	Pass	Pass

Attribution & Acknowledgments

Original Foundation: Jackrong/Qwopus3.8-27B-Flash & Qwen AI
1M YaRN Scaling & Quantization Suite: Solstice-AI

Downloads last month: -

Safetensors

Model size

28B params

Tensor type

U32

BF16

MLX

Hardware compatibility

8-bit

Model tree for Solstice-AI/Qwopus3.8-27B-Flash-mlx-oQ8e-1M

Base model

Qwen/Qwen3.8-27B

Finetuned

Jackrong/Qwopus3.8-27B-Flash

Quantized

(35)

this model