Llama 3 70B

Deprecated

llama3-70b-8192

Try it in Playground

TOKEN SPEED

~330 tps

INPUT

Text

OUTPUT

Text

CAPABILITIES

Tool Use, JSON Mode

PRICING

Input

$0.59

1.7M / $1

Output

$0.79

1.3M / $1

LIMITS

CONTEXT WINDOW

8,192

MAX OUTPUT TOKENS

8,192

Key Technical Specifications

Model Architecture

A 70 billion parameter transformer that includes enhanced attention mechanisms and optimized training objectives. It offers solid instruction-following capabilities and reduced hallucinations.

Performance Metrics

The model demonstrates solid performance across various benchmarks:

MMLU (5-shot): 79.5% accuracy, showing strong general knowledge
GSM-8K (8-shot, CoT): 93.0% accuracy in mathematical reasoning
HumanEval (0-shot): 81.7% pass rate in code generation

Use Cases

Dialogue Applications

Ideal for building reliable conversational experiences with consistent outputs:

Customer support and service chatbots
Interactive assistants and guides
Educational dialogue systems
Conversational interfaces for applications

Content Generation

Excels at creating high-quality content with a balance of creativity and accuracy:

Marketing and promotional content
Documentation and technical writing
Creative writing and storytelling
Content adaptation and summarization

Best Practices

Structure your prompts: Break complex tasks into clear steps for more reliable outputs
Enable JSON mode: For generating structured data and maintaining consistent output formats
Include examples: Add sample outputs or specific formats to guide complex generations

Get Started with llama3-70b

Experience the versatile llama3-70b-8192 with Groq speed now:

shell

pip install groq

Python

from groq import Groq
client = Groq()
completion = client.chat.completions.create(
    model="llama3-70b-8192",
    messages=[
        {
            "role": "user",
            "content": "Explain why fast inference is critical for reasoning models"
        }
    ]
)
print(completion.choices[0].message.content)