Llama 3 70B
Meta · dense · 70.6B parameters · 8,192 context
Parameters
70.6B
Context Window
8K tokens
Architecture
Dense
Best GPU
H200 SXM
Cheapest API
$0.88/M
Quality Score
80/100
Intelligence Brief
Llama 3 70B is a 70.6B parameter DENSE model from Meta, featuring Grouped Query Attention (GQA) with 80 layers and 8,192 hidden dimensions. With a 8,192 token context window, it supports structured output, code, math, multilingual. On standardized benchmarks, it achieves MMLU 82, HumanEval 56, GSM8K 90. The most cost-effective API deployment is via together at $0.88/M output tokens. For self-hosted inference, H200 SXM delivers optimal throughput at $2553/month.
Architecture Details
Memory Requirements
BF16 Weights
141.2 GB
FP8 Weights
70.6 GB
INT4 Weights
35.3 GB
GPU Compatibility Matrix
Llama 3 70B is compatible with 37% of GPU configurations across 41 GPUs at 3 precision levels.
GPU Recommendations
FP8 · 1 GPU · tensorrt-llm
100/100
score
Throughput
560.0 tok/s
Latency (ITL)
1.8ms
Est. TTFT
0ms
Cost/Month
$2553
Cost/M Tokens
$1.73
FP8 · 1 GPU · tensorrt-llm
100/100
score
Throughput
474.0 tok/s
Latency (ITL)
2.1ms
Est. TTFT
0ms
Cost/Month
$940
Cost/M Tokens
$0.75
FP8 · 1 GPU · tensorrt-llm
100/100
score
Throughput
474.0 tok/s
Latency (ITL)
2.1ms
Est. TTFT
0ms
Cost/Month
$2838
Cost/M Tokens
$2.28
Deployment Options
API Deployment
together
$0.88/M
output tokens
Single GPU
H200 SXM
$2553/mo
Min VRAM: 71 GB
Multi-GPU
H100 SXM x2
560.0 tok/s
TP· $3587/mo
API Pricing Comparison
| Provider | Input $/M | Output $/M | Badges |
|---|---|---|---|
| together | $0.88 | $0.88 | Cheapest |
Cost Analysis
| Provider | Input $/M | Output $/M | ~Monthly Cost |
|---|---|---|---|
| togetherBest Value | $0.88 | $0.88 | $9 |
Cost per 1,000 Requests
Short (500 tok)
$0.62
via together
Medium (2K tok)
$2.46
via together
Long (8K tok)
$8.80
via together
Performance Estimates
Throughput by GPU
VRAM Breakdown (H200 SXM, FP8)
Precision Impact
bf16
141.2 GB
weights/GPU
fp8
70.6 GB
weights/GPU
~560.0 tok/s
int4
35.3 GB
weights/GPU
Quality Benchmarks
Capabilities
Features
Supported Frameworks
Supported Precisions
Where to Deploy Llama 3 70B
Self-Hosted Infrastructure
Similar Models
Llama 3 70B 1M Context
70.6B params · dense
Quality: 50
from $1.50/M
DeepSeek R1 Distill 70B
70.6B params · dense
Quality: 88
from $0.88/M
Llama 3.1 70B
70.6B params · dense
Quality: 75
from $0.79/M
Llama 3.3 70B
70.6B params · dense
Quality: 77
from $0.79/M
Hermes 3 70B
70.6B params · dense
Quality: 50
from $0.88/M
Frequently Asked Questions
How much VRAM does Llama 3 70B need for inference?
Llama 3 70B requires approximately 141.2 GB of VRAM at BF16 precision, 70.6 GB at FP8, or 35.3 GB at INT4 quantization. Additional VRAM is needed for KV-cache (327680 bytes per token) and activations (~2.50 GB).
What is the best GPU for Llama 3 70B?
The top recommended GPU for Llama 3 70B is the H200 SXM using FP8 precision. It achieves approximately 560.0 tokens/sec at an estimated cost of $2553/month ($1.73/M tokens). Score: 100/100.
How much does Llama 3 70B inference cost?
Llama 3 70B API inference starts from $0.88/M input tokens and $0.88/M output tokens. Self-hosted inference costs depend on your GPU configuration — use our ROI calculator for a detailed breakdown.