FitMyLLM
Meta/Dense

MetaLlama 3.1 405B

Llama 3.1 405B — largest open-weight model. Rivals GPT-4 on many tasks.

chatcodingreasoningThinkingTool UseDistilled
405B
Parameters
128K
Context length
20
Benchmarks
17
Quantizations
100K
HF downloads
Architecture
Dense
Released
2024-07-23
Layers
126
KV Heads
8
Head Dim
128
Family
llama

Quantization Options

QuantBitsVRAMQuality
IQ2_XXS2.38121.0 GBlow
IQ2_M2.93148.8 GBlow
Q2_K3.16160.5 GBlow
IQ3_XXS3.25165.0 GBlow
IQ3_XS3.5177.7 GBlow
Q3_K_S3.64184.8 GBlow
IQ3_M3.76190.8 GBlow
Q3_K_M4203.0 GBlow
Q3_K_L4.3218.2 GBmoderate
IQ4_XS4.46226.3 GBmoderate
Q4_K_S4.67236.9 GBmoderate
Q4_K_M4.89248.0 GBgood
Q5_K_S5.57282.5 GBgood
Q5_K_M5.7289.1 GBgood
Q6_K6.56332.6 GBexcellent
Q8_08.5430.8 GBlossless
FP1616810.5 GBlossless

Select your GPU above to see speed estimates and compatibility for each quantization.

▸ READY TO RUN THIS?RENT BY THE HOUR

RENT A GPU AND RUN LLAMA 3.1 405B NOW

Rent on RunPod →Or Vast.ai →

Spin up an A100 / H100 / 4090 in ~60s. Pay by the second. Cancel anytime.

Community Ratings

Loading ratings...

Benchmarks (20)

Arena Elo1501
HumanEval89.0
IFEval88.6
BBH88.5
MATH73.8
MMLU-PRO73.3
GPQA Diamond72.7
AIME69.7
AA Math69.7
LiveCodeBench68.6
GPQA51.1
IFBench34.8
SciCode34.6
τ²-Bench26.6
AA Long Context20.0
AA Intelligence18.6
AA Coding16.0
HLE10.3
Terminal-Bench9.8
MATH-5003.0

Run this model

▸Easiest way to get started·Beginners
DOCS ↗
curl -fsSL https://ollama.com/install.sh | sh
$ollama run llama:405b-q4_K_M

Tag may need adjustment — check ollama.com/library/llama for available tags.

▸ SETUP GUIDE
>_

Auto-setup with fitmyllm CLI

Detects your GPU, recommends the best model, downloads it, and starts chatting — zero config. Benchmarks your speed and contributes anonymous data to improve predictions.

pip install fitmyllmthen run fitmyllmLearn more
Auto-detect GPULive tok/s in chatSpeed benchmarks9 inference engines
HuggingFaceGGUF DownloadsBuild Hardware

GPUs that can run this model

At Q4_K_M quantization. Sorted by minimum VRAM.

AMD Radeon Instinct MI325X
288 GB VRAM • 10300 GB/s
AMD
$20000
AMD Radeon Instinct MI350X
288 GB VRAM • 8190 GB/s
AMD
$25000
AMD Radeon Instinct MI355X
288 GB VRAM • 8190 GB/s
AMD
$30000
Apple M4 Ultra (384GB)
384 GB VRAM • 1092 GB/s
APPLE
$9999
Apple M5 Ultra (384GB)
384 GB VRAM • 1228 GB/s
APPLE

Find the best GPU for Llama 3.1 405B

Build Hardware for Llama 3.1 405B

Model Card

View on HuggingFace

Llama 3.1 405B — largest open-weight model. Rivals GPT-4 on many tasks.

▸ SPEC SHEET

Llama 3.1 405B — 405B Dense.

▸ SPECIFICATIONS
PARAMETERS
405B
ARCHITECTURE
Dense Transformer
CONTEXT LENGTH
128K tokens
CAPABILITIES
chat, coding, reasoning
RELEASE DATE
2024-07-23
PROVIDER
Meta
FAMILY
llama
▸ VRAM REQUIREMENTS
QUANTBPWVRAMQUALITY
IQ2_XXS2.38121.0 GB65%
IQ2_M2.93148.8 GB75%
Q2_K3.16160.5 GB78%
IQ3_XXS3.25165.0 GB82%
IQ3_XS3.5177.7 GB84%
Q3_K_S3.64184.8 GB85%
IQ3_M3.76190.8 GB86%
Q3_K_M4203.0 GB88%
Q3_K_L4.3218.2 GB90%
IQ4_XS4.46226.3 GB92%
Q4_K_S4.67236.9 GB93%
Q4_K_M4.89248.0 GB94%
Q5_K_S5.57282.5 GB96%
Q5_K_M5.7289.1 GB96%
Q6_K6.56332.6 GB97%
Q8_08.5430.8 GB100%
FP1616810.5 GB100%
§ 01BENCHMARK SCORES
HumanEval89.0
MMLU-PRO73.3
MATH73.8
IFEval88.6
BBH88.5
GPQA51.1
Arena Elo1501.0
GPQA Diamond72.7
LiveCodeBench68.6
AIME69.7
HLE10.3
AA Intelligence18.6
AA Coding16.0
AA Math69.7
aa_ifbench34.8
aa_terminal_bench9.8
aa_tau226.6
aa_scicode34.6
aa_lcr20.0
MATH-5003.0
§ 03COMPATIBLE GPUs
5 @ Q4_K_M
AMD Radeon Instinct MI325X
288 GB · 10300 GB/s
AMD Radeon Instinct MI350X
288 GB · 8190 GB/s
AMD Radeon Instinct MI355X
288 GB · 8190 GB/s
Apple M4 Ultra (384GB)
384 GB · 1092 GB/s
Apple M5 Ultra (384GB)
384 GB · 1228 GB/s