Llama-3.1-8B-Instruct vs Hermes-3-Llama-3.1-8B
Specs, VRAM requirements and download trends — updated 6 August 2026.
NousResearch · B
Params
8.0 B
Context
131,072
30d
558 K
llama3 licence · conditional
Specification comparison
Differences are highlighted; identical values are muted.
| Specification | Llama-3.1-8B-Instruct | Hermes-3-Llama-3.1-8B |
|---|---|---|
| Parameters | 8.0 B | 8.0 B |
| Architecture | LlamaForCausalLM | LlamaForCausalLM |
| Context length | — | 131,072 |
| Licence | llama3.1 | llama3 |
| Commercial use | Conditional | Conditional |
| Languages | 8 | 1 |
| Downloads 30d | 8,014,793 | 558,143 |
| Downloads all time | 175.3 M | 4.1 M |
| Quantizations on the Hub | SAFETENSORS | SAFETENSORS |
| Gated | Yes — accept licence | No |
| First seen on the Hub | 2024-07-18 | 2024-07-28 |
Download trend
Daily snapshots, last 10 days (28 Jul – 6 Aug)
8.0 M515 K
Llama-3.1-8B-Instruct
Hermes-3-Llama-3.1-8B
VRAM side by side
On RTX 4090 · 24 GB · 8K context unless noted
Llama-3.1-8B-Instruct · fp16
19.4 GB / 24 GB
Hermes-3-Llama-3.1-8B · fp16 @ 128K ctx
37.4 GB / 24 GB