Hermes 4.3 36B — Hardware Requirements & GPU Compatibility
ChatReasoningRoleplay
5.7K downloads
137 likesSpecifications
- Publisher
- Nous Research
- Family
- Hermes
- Parameters
- 36B
- Architecture
- SeedOssForCausalLM
- Context Length
- 524,288 tokens
- Vocabulary Size
- 155,136
- Release Date
- 2025-12-06
- License
- Apache 2.0
Get Started
HuggingFace
How Much VRAM Does Hermes 4.3 36B Need?
Select a quantization to see compatible GPUs below.
| Quantization | Bits | VRAM | + Context | File Size | Quality |
|---|---|---|---|---|---|
| BF16 | 16.00 | 72.6 GB | 141.0 GB | 72.00 GB | Brain floating point 16 — preferred for training |
Which GPUs Can Run Hermes 4.3 36B?
BF16 · 72.6 GBHermes 4.3 36B (BF16) requires 72.6 GB of VRAM to load the model weights. For comfortable inference with headroom for KV cache and system overhead, 95+ GB is recommended. Using the full 524K context window can add up to 68.5 GB, bringing total usage to 141.0 GB. No single GPU has enough memory — multi-GPU or cluster setups are needed.
Which Devices Can Run Hermes 4.3 36B?
BF16 · 72.6 GB5 devices with unified memory can run Hermes 4.3 36B, including NVIDIA DGX H100, NVIDIA DGX A100 640GB.
Related Models
Frequently Asked Questions
- How much VRAM does Hermes 4.3 36B need?
Hermes 4.3 36B requires 72.6 GB of VRAM at BF16. Full 524K context adds up to 68.5 GB (141.0 GB total).
VRAM = Weights + KV Cache + Overhead
Weights = 36B × 16 bits ÷ 8 = 72 GB
KV Cache + Overhead ≈ 0.6 GB (at 2K context + ~0.3 GB framework)
KV Cache + Overhead ≈ 69 GB (at full 524K context)
VRAM usage by quantization
BF1672.6 GBBF16 + full context141.0 GB- Can NVIDIA GeForce RTX 5090 run Hermes 4.3 36B?
No — Hermes 4.3 36B requires at least 72.6 GB at BF16, which exceeds the NVIDIA GeForce RTX 5090's 32 GB of VRAM.
- Can I run Hermes 4.3 36B on a Mac?
Hermes 4.3 36B requires at least 72.6 GB at BF16, which exceeds the unified memory of most consumer Macs. You would need a Mac Studio or Mac Pro with a high-memory configuration.
- Can I run Hermes 4.3 36B locally?
Yes — Hermes 4.3 36B can run locally on consumer hardware. At BF16 quantization it needs 72.6 GB of VRAM. Popular tools include Ollama, LM Studio, and llama.cpp.
- How fast is Hermes 4.3 36B?
At BF16, Hermes 4.3 36B can reach ~40 tok/s on AMD Instinct MI300X. Speed depends mainly on GPU memory bandwidth. Real-world results typically within ±20%.
tok/s = (bandwidth GB/s ÷ model GB) × efficiency
Example: AMD Instinct MI300X → 5300 ÷ 72.6 × 0.55 = ~40 tok/s
Estimated speed at BF16 (72.6 GB)
AMD Instinct MI300X~40 tok/sNVIDIA H100 SXM~30 tok/sAMD Instinct MI250X~25 tok/sReal-world results typically within ±20%. Speed depends on batch size, quantization kernel, and software stack.
- What's the download size of Hermes 4.3 36B?
At BF16, the download is about 72.00 GB.
- Which GPUs can run Hermes 4.3 36B?
No single consumer GPU has enough VRAM to run Hermes 4.3 36B at BF16 (72.6 GB). Multi-GPU or professional hardware is required.
- Which devices can run Hermes 4.3 36B?
5 devices with unified memory can run Hermes 4.3 36B at BF16 (72.6 GB), including Mac Pro M2 Ultra (192 GB), Mac Studio M2 Ultra (192 GB), Mac Studio M4 Max (128 GB), NVIDIA DGX A100 640GB. Apple Silicon Macs use unified memory shared between CPU and GPU, making them well-suited for local LLM inference.