---
title: BTL-4-OptiQ-4bit
canonical_url: "https://www.modelscope.cn/models/mlx-community/BTL-4-OptiQ-4bit"
md_url: "https://www.modelscope.cn/models/mlx-community/BTL-4-OptiQ-4bit.md"
repository: mlx-community/BTL-4-OptiQ-4bit
last_updated: 2026-09-15
license: apache-2.0
pipeline_tag: text-generation
tasks:
  - text-generation
model_type:
  - qwen3_5_moe
architectures:
  - Qwen3_5MoeForConditionalGeneration
base_model:
  - badtheorylabs/BTL-4
base_model_relation: quantized
parameters: 6.1B
tensor_type:
  - U32
  - BF16
library_name:
  - mlx
  - safetensors
  - pytorch
frameworks:
  - pytorch
downloads: 25
stars: 0
tags:
  - mlx
  - optiq
  - quantized
  - 4bit
  - mixed-precision
  - moe
  - agentic
  - tool-use
  - code
  - apple-silicon
---

# BTL-4-OptiQ-4bit

> BTL-4-OptiQ-4bit - mlx-community 在 ModelScope 开源的模型。mlx-community/BTL-4-OptiQ-4bit

mlx-community/BTL-4-OptiQ-4bit 是 ModelScope 魔搭社区上的 6.1B 参数text-generation模型，采用 apache-2.0 许可，基于 badtheorylabs/BTL-4 构建。

- **Repository**: mlx-community/BTL-4-OptiQ-4bit
- **License**: apache-2.0
- **Tasks**: text-generation
- **Parameters**: 6.1B
- **Base model**: badtheorylabs/BTL-4
- **Tags**: mlx, optiq, quantized, 4bit, mixed-precision, moe, agentic, tool-use, code, apple-silicon
- **Downloads**: 25
- **Stars**: 0
- **Last updated**: 2026-09-15

Source: https://www.modelscope.cn/models/mlx-community/BTL-4-OptiQ-4bit

---

# mlx-community/BTL-4-OptiQ-4bit

> **Built with [mlx-optiq](https://mlx-optiq.com)**, the MLX-native toolkit to quantize, fine-tune, and serve LLMs locally on Apple Silicon, no PyTorch and no cloud. [All OptiQ quants](https://mlx-optiq.com/models) · [Docs](https://mlx-optiq.com/docs/) · [Qwen3.5 family](https://mlx-optiq.com/docs/qwen3.5)

An OptiQ mixed-precision quant of [badtheorylabs/BTL-4](https://huggingface.co/badtheorylabs/BTL-4), a 35B agentic reasoning model built for tool use, software engineering and long-horizon agent work. 22.2 GB on disk, down from 70.2 GB at bf16.

## What it is

| Property | Value |
|---|---|
| Base | [badtheorylabs/BTL-4](https://huggingface.co/badtheorylabs/BTL-4) |
| Architecture | `qwen3_5_moe` — sparse mixture-of-experts |
| Method | OptiQ mixed-precision, per-layer bit allocation reused from the base family |
| On disk | 22.2 GB (bf16: 70.2 GB) |

BTL-4 keeps the architecture of the family it is derived from, so which layers tolerate fewer bits is unchanged and a fresh sensitivity sweep would only rediscover the same answer. The per-layer allocation comes from [Qwen3.5-35B-A3B-OptiQ-4bit](https://huggingface.co/mlx-community/Qwen3.5-35B-A3B-OptiQ-4bit): 512 of 512 layers matched, with the routed experts mostly at 4-bit and attention, router and layer edges kept at 8-bit.

No Capability Score is published for this quant. The base model's own benchmarks are on [its card](https://huggingface.co/badtheorylabs/BTL-4).

## Run it

```bash
pip install mlx-optiq
optiq serve --model mlx-community/BTL-4-OptiQ-4bit
```

That gives you an OpenAI and Anthropic compatible endpoint with mixed-precision KV cache, tool-call healing and prompt caching — useful for an agentic model, where a malformed tool call costs a whole turn.

Or from Python:

```python
from mlx_lm import load, generate

model, tokenizer = load("mlx-community/BTL-4-OptiQ-4bit")
prompt = tokenizer.apply_chat_template(
    [{"role": "user", "content": "List the files in the current directory."}],
    add_generation_prompt=True, tokenize=False,
)
print(generate(model, tokenizer, prompt=prompt, max_tokens=512))
```

## Links

- **Project website:** [mlx-optiq.com](https://mlx-optiq.com/)
- **All OptiQ quants:** [mlx-optiq.com/models](https://mlx-optiq.com/models)
- **PyPI:** [pypi.org/project/mlx-optiq](https://pypi.org/project/mlx-optiq/)
- **Base model:** [badtheorylabs/BTL-4](https://huggingface.co/badtheorylabs/BTL-4)