---
license: mit
pipeline_tag: text-generation
library_name: mlx
base_model: deepseek-ai/DeepSeek-R1-Distill-Llama-8B
base_model_relation: quantized
quantized_by: Outlier-Ai
tags:
- 4-bit
- 4bit
- apple-silicon
- chain-of-thought
- chat
- conversational
- deepseek
- deepseek-r1
- deepseek-r1-distill
- edge-ai
- instruct
- local-llm
- m1
- m2
- m3
- m4
- mac
- mac-mini
- mac-studio
- macbook-air
- macbook-pro
- macos
- metal
- mlx
- mlx-community
- mlx-lm
- no-cloud
- offline
- on-device
- outlier
- outlier-app
- private
- quantized
- r1
- r1-distill
- reasoning
- safetensors
- text-generation
- thinking
language:
- en
- zh
- fr
- es
- pt
- de
- ru
- ja
- ko
- ar
widget:
- example_title: Widget math
  messages:
  - role: user
    content: If it takes 5 machines 5 minutes to make 5 widgets, how long would it
      take 100 machines to make 100 widgets? Show reasoning.
- example_title: Logic grid
  messages:
  - role: user
    content: Alice, Bob, and Carol each own a different pet. Alice does not own the
      dog. Bob owns the parrot. Who owns what?
- example_title: Word problem
  messages:
  - role: user
    content: A train leaves station A at 60 mph and another leaves station B, 180
      miles away, at 40 mph toward each other. When do they meet?
---

# DeepSeek-R1-Distill-Llama-8B — MLX 4-bit

4-bit MLX conversion of [DeepSeek-R1-Distill-Llama-8B](https://huggingface.co/deepseek-ai/DeepSeek-R1-Distill-Llama-8B) for Apple Silicon. Runs natively on
M1, M2, M3, and M4 Macs via `mlx_lm`. Faithful port of the upstream weights — no
fine-tuning, no merge.

## Quick facts

- **Upstream base:** [DeepSeek-R1-Distill-Llama-8B](https://huggingface.co/deepseek-ai/DeepSeek-R1-Distill-Llama-8B)
- **Format:** MLX 4-bit (`mlx_lm.convert --q-bits 4 --q-group-size 64`)
- **License:** `mit`
- **Use case:** reasoning
- **Runtime:** mlx-lm, LM Studio, Jan, Outlier desktop app
- **Publisher:** Outlier-Ai (solo, Mac-native AI platform)

## Quickstart (mlx-lm)

```bash
pip install -U mlx-lm
python -m mlx_lm.generate \
  --model Outlier-Ai/DeepSeek-R1-Distill-Llama-8B-MLX-4bit \
  --prompt "Write a haiku about compilers." \
  --max-tokens 256
```

## Quickstart (LM Studio)

Search for `Outlier-Ai/DeepSeek-R1-Distill-Llama-8B-MLX-4bit`, then Load and chat.

## Quickstart (Outlier desktop app)

One-click install from [outlier.host](https://outlier.host). Listed in the library
picker under `reasoning`.

## Why this repo exists

This is a canonically-named twin of [Outlier-Ai/Outlier-R1-Distill-Llama-8B-MLX-4bit](https://huggingface.co/Outlier-Ai/Outlier-R1-Distill-Llama-8B-MLX-4bit).
Same weights, same quant, different name. The branded repo carries the Outlier
identity; this one carries the upstream identity so it's findable via searches
like "deepseek-r1-distill-llama-8b 4bit mlx". Use whichever path you came across — both
point to the same files.

## Provenance

- Converted via `mlx_lm.convert --q-bits 4 --q-group-size 64`
- Source: [DeepSeek-R1-Distill-Llama-8B](https://huggingface.co/deepseek-ai/DeepSeek-R1-Distill-Llama-8B)
- Date produced: 2026-04-19 (Day 19)
- Conversion factory: Outlier MEGA-INFINITE phase3

## Related

- **Branded twin (same weights):** [Outlier-Ai/Outlier-R1-Distill-Llama-8B-MLX-4bit](https://huggingface.co/Outlier-Ai/Outlier-R1-Distill-Llama-8B-MLX-4bit)
- **Outlier org on HuggingFace:** [Outlier-Ai](https://huggingface.co/Outlier-Ai)
- **Desktop app:** [outlier.host](https://outlier.host)
- **Founders lifetime ($200, 500-seat cap):** [buy.polar.sh](https://buy.polar.sh/polar_cl_mJfYZsEpEMDcYrgxzvTdnahSeSQNq1UYLqV0l08CUhW)
- **Discord:** [discord.gg/Hapennmdn9](https://discord.gg/Hapennmdn9)

## Collections

- **Outlier for Apple Silicon (MLX):** [huggingface.co/collections/Outlier-Ai/outlier-consumer-edition-69e2fb4a0df119ea1747275e](https://huggingface.co/collections/Outlier-Ai/outlier-consumer-edition-69e2fb4a0df119ea1747275e)
- **Outlier Research MoE:** [huggingface.co/collections/Outlier-Ai/outlier-research-69e2fb3a71984614b3c7a279](https://huggingface.co/collections/Outlier-Ai/outlier-research-69e2fb3a71984614b3c7a279)
- **Outlier Server V3.2:** [huggingface.co/collections/Outlier-Ai/outlier-server-v32-69e2fb4b71984614b3c7a4a3](https://huggingface.co/collections/Outlier-Ai/outlier-server-v32-69e2fb4b71984614b3c7a4a3)

## Attribution

Base model by the upstream author (see [DeepSeek-R1-Distill-Llama-8B](https://huggingface.co/deepseek-ai/DeepSeek-R1-Distill-Llama-8B)). Outlier contributes
the MLX 4-bit conversion and desktop-app packaging. Capability credit for the base
belongs entirely to upstream.

## License

Redistributed under the upstream license: `mit`. Respect the terms of
[DeepSeek-R1-Distill-Llama-8B](https://huggingface.co/deepseek-ai/DeepSeek-R1-Distill-Llama-8B) when using this conversion.