Q: What quantization of Hermes 3 Llama 3.1 70B should I use on a NVIDIA GeForce RTX 5090?

For 32 GB VRAM on the NVIDIA GeForce RTX 5090, the Q4_K_M variant is the best fit. Estimated ~11 tokens/sec on the Q4_K_M quantization.

Q: How fast does Hermes 3 Llama 3.1 70B run on NVIDIA GeForce RTX 5090?

Roughly 11 tokens/sec for Q4_K_M. Real speed depends on context length, backend (Ollama, llama.cpp, LM Studio), and KV cache size.

Q: What if NVIDIA GeForce RTX 5090 is not enough for Hermes 3 Llama 3.1 70B?

Consider upgrading to Apple M4 Pro (48 GB VRAM) which fits the recommended 48 GB target. Or pick a smaller quantization to stay on your current card.

Question 1

Can I run Hermes 3 Llama 3.1 70B on a NVIDIA GeForce RTX 5090?

Accepted Answer

Sort of — NVIDIA GeForce RTX 5090 can run Hermes 3 Llama 3.1 70B (Q4_K_M) only by spilling layers to RAM. Generation will be slow. CPU + GPU hybrid — not enough VRAM (32 GB < 42 GB min), but 64 GB RAM is sufficient. Expect significantly slower inference.

Question 2

What quantization of Hermes 3 Llama 3.1 70B should I use on a NVIDIA GeForce RTX 5090?

Accepted Answer

For 32 GB VRAM on the NVIDIA GeForce RTX 5090, the Q4_K_M variant is the best fit. Estimated ~11 tokens/sec on the Q4_K_M quantization.

Question 3

How fast does Hermes 3 Llama 3.1 70B run on NVIDIA GeForce RTX 5090?

Accepted Answer

Roughly 11 tokens/sec for Q4_K_M. Real speed depends on context length, backend (Ollama, llama.cpp, LM Studio), and KV cache size.

Question 4

What if NVIDIA GeForce RTX 5090 is not enough for Hermes 3 Llama 3.1 70B?

Accepted Answer

Consider upgrading to Apple M4 Pro (48 GB VRAM) which fits the recommended 48 GB target. Or pick a smaller quantization to stay on your current card.

Can I Run Hermes 3 Llama 3.1 70B on NVIDIA GeForce RTX 5090?

Share this matchup

Every Hermes 3 Llama 3.1 70B quantization on NVIDIA GeForce RTX 5090

Upgrade options that fit Hermes 3 Llama 3.1 70B better