mlx-community / text-generation updated 1 year ago

Llama-3.1-8B-Instruct-4bit

Name: Llama-3.1-8B-Instruct-4bit
Author: mlx-community

The Model mlx-community/Llama-3.1-8B-Instruct-4bit was converted to MLX format from meta-llama/Llama-3.1-8B-Instruct using mlx-lm version 0.21.4.

Params

1.3 B

Context

131,072

Downloads 30d

905 K

Likes

Commercial use: conditional · llama3.1 Not gated SAFETENSORS 8 languages View on Hugging Face ↗

Download history

daily snapshots · 9 days

905 K185 K

Sep 1Sep 4Sep 7Sep 9

Can you run it?

Estimated VRAM at 8K context unless noted. Pick your hardware to see the verdict per quantization.

File	Quant	Size	Est. VRAM	Verdict on RTX 4090 · 24 GB
model.safetensors	u32	4.5 GB	5.7 GB	✅ Runs comfortably
model.safetensors (bf16, full)	bf16 + 128K ctx	4.5 GB	8.5 GB	✅ Runs comfortably

Estimate: file size × 1.1 + KV cache at 8K + 0.5 GB overhead. Not a benchmark — how we calculate this.

Run it

copy-paste, exact tags checked against the Hub

~ · curl · api/v1

$ curl -s https://aimodelscomparison.com/api/v1/models/llama-3-1-8b-instruct-4bit
{
  "hf_id": "mlx-community/Llama-3.1-8B-Instruct-4bit",
  "params_b": 1.25,
  "context_length": 131072,
  "license": { "id": "llama3.1", "commercial": "conditional" },
  "downloads_30d": 904691,
  "vram_estimates": [
    { "quant": "u32", "gb": 5.7 }
  ],
  "updated_at": "2026-09-01T01:00:33Z"
}

est. VRAM —on RTX 4090 · 24 GBJSON API →

Specifications

Architecture: LlamaForCausalLM
Parameters: 1.3 B
Tensor type: U32
Context length: 131,072
Vocabulary: 128,256
Layers / heads: 32 / 32
Licence: llama3.1
First seen on the Hub: 2025-02-15
Base model: Llama-3.1-8B-Instruct
Training datasets: undisclosed
Added to our catalog: 2026-09-01

Family

Base model and the most-downloaded derivatives in the catalog.

Base Llama-3.1-8B-Instruct meta-llama 5.6 M

Compare with

Qwen3-0.6B 21.5 M Qwen2.5-1.5B-Instruct 7.3 M Qwen2.5-0.5B-Instruct 6.4 M Llama-3.2-1B-Instruct 6.1 M Qwen2.5-3B-Instruct 6.0 M Qwen3-1.7B 3.4 M

Appears in

Best open coding models Commercial-use text models Models that fit in 12 GB