Meta-Llama-3-8B-Instruct vs Qwen3Guard-Gen-4B
Specs, VRAM requirements and download trends — updated 21 August 2026.
meta-llama · A · Open weights
Params
8.0 B
Context
—
30d
386 K
llama3 licence · conditional
Qwen · B · Open weights
Params
4.4 B
Context
32,768
30d
207 K
apache-2.0 · commercial OK
Specification comparison
Differences are highlighted; identical values are muted.
| Specification | Meta-Llama-3-8B-Instruct | Qwen3Guard-Gen-4B |
|---|---|---|
| Parameters | 8.0 B | 4.4 B |
| Architecture | LlamaForCausalLM | Qwen3ForCausalLM |
| Context length | — | 32,768 |
| Licence | llama3 | apache-2.0 |
| Commercial use | Conditional | Allowed |
| Languages | 1 | — |
| Downloads 30d | 385,543 | 206,801 |
| Downloads all time | 49.1 M | 780 K |
| Quantizations on the Hub | SAFETENSORS | SAFETENSORS |
| Gated | Yes — accept licence | No |
| First seen on the Hub | 2024-04-17 | 2025-09-23 |
Download trend
Daily snapshots, last 90 days (13 Aug – 10 Nov)
2.0 M207 K
Meta-Llama-3-8B-Instruct
Qwen3Guard-Gen-4B
VRAM side by side
On RTX 4090 · 24 GB · 8K context unless noted
Qwen3Guard-Gen-4B · fp16
10.9 GB / 24 GB
Meta-Llama-3-8B-Instruct · fp16
19.4 GB / 24 GB