This model is derived from mlx-community/Huihui-Qwen3.5-9B-Claude-4.6-Opus-abliterated-4bit. The original model works with mlx-vlm but had issues with mlx-swift-lm:

❌ Text-only chat template (no image support)

What was fixed

Issue	Fix
chat_template.jinja	Replaced with VLM-aware template handling image/video content arrays

Usage with mlx-swift-lm

swift

let modelDir = URL(fileURLWithPath: "/path/to/model")
let container = try await VLMModelFactory.shared.loadContainer(
    from: modelDir,
    using: LocalTokenizerLoader()
)
let session = ChatSession(container)
let response = try await session.respond(
    to: "Describe the image",
    image: .url(URL(fileURLWithPath: "/path/to/image.jpg"))
)

Download

To get the complete model, download the weight files from the original repo (mlx-community/Huihui-Qwen3.5-9B-Claude-4.6-Opus-abliterated-4bit), then replace the config files with the fixed versions from this repo.

Fixed for mlx-swift-lm by monyschuk

monyschuk/Qwen3.5-9B-Claude-Opus-abliterated-VLM-MLX

Qwen3.5 9B Claude Opus Abliterated — MLX VLM

About

What was fixed

Usage with mlx-swift-lm

Download

monyschuk/
Qwen3.5-9B-Claude-Opus-abliterated-VLM-MLX