---
license: apache-2.0
language:
- en
base_model:
- Qwen/Qwen3.6-27B
tags:
- MLX
- mtplx
- qwen
- fp16
---

# ThinkingCap-Qwen3.6-27B-8bit-FP16-MTPLX

MTPLX-branded multi-token-prediction model for Apple Silicon (MLX).
Forged with [MTPLX Forge](https://github.com/youssofal/MTPLX) from
`bottlecapai/ThinkingCap-Qwen3.6-27B`.

## Verification

- Best depth: D3
- Multiplier vs autoregressive baseline: 1.36×
- Verified on: Apple M2 Max
- Sampler: temperature 0.6 · top_p 0.95 · top_k 20

See `mtplx_runtime.json` for the full verification record.

## Usage

```bash
# MTPLX picks this model up automatically when downloaded:
mtplx pull <owner>/ThinkingCap-Qwen3.6-27B-8bit-FP16-MTPLX
mtplx start chat
```

## License

See LICENSE.