LLMBOTTLENECK.COM
NousResearch / llama

Hermes-3-Llama-3.1-405B

405.9 billion parameters.

Open in the calculator →

Architecturepublished data
Quick answer
Hermes-3-Llama-3.1-405B needs about 248.1 GB at Q4_K_M — the format most people download — with 8,192 tokens of context. None of the 18 common devices listed below holds it entirely at this setting. No hardware for it? Rent a GPU that runs it, priced live.

Sized at 8,192 tokens of context with the whole model in device memory, across 18 common devices. Speeds are estimates, not benchmarks.

Weight-file size by format

FormatSizeRangeBasis
FP16811.7 GB807.5 GB – 815.8 GBreconstructed size
Q8_0431.2 GB429.0 GB – 433.4 GBreconstructed size
Q6_K332.9 GB331.2 GB – 334.6 GBreconstructed size
Q5_K_M286.6 GB284.9 GB – 288.1 GBreconstructed size
Q5_0279.3 GB277.6 GB – 280.7 GBreconstructed size
Q4_K_M243.1 GB241.3 GB – 244.3 GBreconstructed size
Q4_0228.9 GB227.1 GB – 230.0 GBreconstructed size
Q3_K_M197.0 GB183.2 GB – 210.8 GBreconstructed size
Q2_K145.6 GB139.8 GB – 151.4 GBreconstructed size

A file size is not the memory a run needs: the KV cache and the runtime reserve come on top, and the calculator adds both.

What the labels mean
published data
Read from a published source — a file's byte count, a model's configuration or a manufacturer's specification — or exact arithmetic on such values. Not a measurement on a machine.
reconstructed size
Weight size reconstructed from the pinned architecture, because no published file exists.
size range
Only a lower and an upper bound are claimed for this weight size.

Hardware ladder

Q4_K_M at 8,192 tokens · 0 of 18 devices hold it entirely
DeviceMemoryVerdictDecodeCalculator
NVIDIA GeForce RTX 5070248.1 GB of 12 GB123 layers on system RAM~0.3 tok/s with offloadOpen →
AMD Radeon™ RX 9070 XT248.1 GB of 16 GB121 layers on system RAM~0.3 tok/s with offloadOpen →
NVIDIA GeForce RTX 4080248.1 GB of 16 GB121 layers on system RAM~0.3 tok/s with offloadOpen →
NVIDIA GeForce RTX 5080248.1 GB of 16 GB121 layers on system RAM~0.3 tok/s with offloadOpen →
NVIDIA GeForce RTX 5070 Ti248.1 GB of 16 GB121 layers on system RAM~0.3 tok/s with offloadOpen →
NVIDIA GeForce RTX 5060 Ti 16GB248.1 GB of 16 GB121 layers on system RAM~0.3 tok/s with offloadOpen →
NVIDIA GeForce RTX 4090248.1 GB of 24 GB117 layers on system RAM~0.3 tok/s with offloadOpen →
AMD Radeon™ RX 7900 XTX248.1 GB of 24 GB117 layers on system RAM~0.3 tok/s with offloadOpen →
NVIDIA GeForce RTX 3090248.1 GB of 24 GB117 layers on system RAM~0.3 tok/s with offloadOpen →
NVIDIA GeForce RTX 5090248.1 GB of 32 GB113 layers on system RAM~0.3 tok/s with offloadOpen →
NVIDIA RTX 6000 Ada Generation248.1 GB of 48 GB104 layers on system RAM~0.3 tok/s with offloadOpen →
Apple M4 Pro248.1 GB of 64 GBdoes not fitnot calibratedOpen →
Apple M5 Pro248.1 GB of 64 GBdoes not fitnot calibratedOpen →
AMD Ryzen AI Max+ 395 with Radeon 8060S248.1 GB of 128 GBdoes not fitnot calibratedOpen →
Apple M3 Max248.1 GB of 128 GBdoes not fitnot calibratedOpen →
Apple M4 Max248.1 GB of 128 GBdoes not fitnot calibratedOpen →
NVIDIA DGX Spark248.1 GB of 128 GBdoes not fitnot calibratedOpen →
Apple M2 Ultra248.1 GB of 192 GBdoes not fitnot calibratedOpen →

Decode is a calibrated estimate from the published calibration; capacity is the manufacturer's published ceiling, not guaranteed free memory. Fitting in memory is not the same as loading: whether the runtime and version you have supports this architecture and format on that machine has not been tested here. Offloaded rows assume there is enough system RAM for the overflow — the calculator checks that against the RAM you declare. Each “Open” link keeps this model, format and context.

Run it in the cloud

Rent a GPU that runs Hermes-3-Llama-3.1-405B

Every rentable machine that holds the whole model, from Vast.ai and RunPod, with this site's speed estimate for each and the provider's own price, read live. Cheapest first; sort by cost per token or speed instead, or switch the format.

Prices · 18:30 UTC, 22 Sept
7 configurations in stock · 9 out of stock
MachineSpeedPer hourPer M tokensWhereRent
8× L40S384 GBcheapestbest value~19 tok/sAbout reading pace$3.73$54.32Vast.aiMarketplace · 99.9% reliableRent on Vast.ai
8× RTX PRO 5000384 GBrecommended~29 tok/sAbout reading pace$6.81$63.66Vast.aiMarketplace · 98.3% reliableRent on Vast.ai
2× H200 NVL282 GB~26 tok/sAbout reading pace$7.20$75.49Vast.aiMarketplace · 98.7% reliableRent on Vast.ai
1× B300270 GB~21 tok/sAbout reading pace$7.89$102.90RunPodSecure CloudRent on RunPod
4× H100 SXM320 GBfastest~37 tok/sFaster than you read$9.07$68.08Vast.aiMarketplace · 99.9% reliableRunPod $13.96/h Rent on Vast.ai
2× H200282 GB~26 tok/sAbout reading pace$9.18$96.23RunPodSecure CloudVast.ai $10.00/h Rent on RunPod
4× H100 PCIe320 GB~22 tok/sAbout reading pace$9.59$120.48Vast.aiMarketplace · 98.3% reliableRent on Vast.ai
No GPU at all

Use Hermes-3-Llama-3.1-405B by the token

For one person chatting, that is about 54× cheaper than the best-value rented card above ($54.32 per million tokens). A rented GPU pays off when you keep it busy — many requests batched together, long agent runs — or when the data must stay on a machine you control, or the exact file you want is not served anywhere.

First time renting a GPU? How it works, in four steps
  1. Create an account and add credit. Both providers are prepaid: $10 is enough to try any card on this page for hours. A new RunPod account that arrives through a referral link gets a one-time $5 credit when it first adds $10.
  2. Pick the card and the count shown here, and a template: llama.cpp or vLLM for an OpenAI-compatible API, or one with a web chat if you only want to talk to the model. “How to launch it” below gives the exact image and command.
  3. Wait for the download. The weights are fetched on the machine; tens of gigabytes take a few minutes on a data-centre connection. You pay by the second from the moment it starts.
  4. Stop it when you are done. A running machine bills even when idle. A stopped one costs nothing per hour, though its disk is usually still billed until you delete it.
How to launch it

No Q4_K_M file of Hermes-3-Llama-3.1-405B is catalogued here, so there is no exact command to vouch for. Start the machine from the llama.cpp server image ghcr.io/ggml-org/llama.cpp:server-cuda and point -hf at a Q4_K_M build from the publisher or a quantizer you trust — search Hugging Face for one. Check its file size against the memory figure on this page before you rent.

Every machine holds the whole model at Q4_K_M and 8,192 tokens of context, no offload. Speeds are this site’s single-stream decode estimates; prices are what each provider’s own API quoted, on-demand, for the whole machine. Vast hosts below 98% measured reliability are left out.

Referral links Vast.ai, RunPod and Novita pay us a share of what you spend if you sign up through these buttons. It costs you nothing, and it never decides an order or a recommendation: both are computed from the live price and the speed, and options that pay us nothing are listed and recommended on the same terms. How we rank

Buying a card instead, or paying by the token? Compare a year of Hermes-3-Llama-3.1-405B three ways →

Under the hood

Architecture, read from the publisher’s file

The numbers every figure above is computed from, with the file they came from.

Architecture

✓ Architecture read from the published config.json

Retrieved 2026-09-01 at pinned commit b9fe8ede1f50; parameter count from the safetensors index at the same revision.

Architecture
llama
Layers
126
Hidden size
16,384
Attention heads
128
KV heads
8
Feed-forward width
53,248
Vocabulary
128,256
Context ceiling
131,072
RoPE theta
500,000

Where the memory goes

tokenembedding× 126 decoder blocksGrouped-query attention128 query · 8 KV headsfull contextFeed-forwardone networkall activeoutputprojectiongrows with contextfixed per token

Grouped-query attention shares each key/value head across 16 query heads, so the KV cache is 6% of what multi-head attention would need at the same context.

Open the exact official config.json ↗

More from NousResearch

Hermes-3-Llama-3.1-70BHermes-3-Llama-3.1-8BHermes-4-405BHermes-4-70BLlama-2-7b-hfMeta-Llama-3.1-8B-Instruct

Use this answer in your own product

As JSON, from the API
curl -s https://llmbottleneck.com/v1/analyze \
  -H "Authorization: Bearer $LLMB_KEY" -H "content-type: application/json" \
  -d '{"model":"nousresearch-hermes-3-llama-3-1-405b","quantization":"Q4_K_M","context":8192,"hardware":"nvidia-geforce-rtx-4090"}'

Same engine, same evidence, every field sourced. Free key in one step →

On your page, as a widget
<script src="https://llmbottleneck.com/widget.js"
  data-model="nousresearch-hermes-3-llama-3-1-405b" data-quantization="Q4_K_M"
  data-hardware="nvidia-geforce-rtx-4090" data-context="8192"></script>

No key needed. Unbranded, with your own buy button, on Pro and Business →

Citing this page

LLM Bottleneck. “Hermes-3-Llama-3.1-405B VRAM and hardware requirements.” Architecture from NousResearch/Hermes-3-Llama-3.1-405B at revision b9fe8ede1f50, retrieved 2026-09-01. https://llmbottleneck.com/models/nousresearch-hermes-3-llama-3-1-405b

Every figure above is either the published value or a reconstruction whose measured error is on the accuracy page.

llmbottleneck
catalogue 2026-10-03models 327devices 135