Perplexity

Perplexity is an AI company focused on providing high-performance language model services. The company offers several advanced models with Function Calling support, primarily including:

• Llama 3.1 70B: The most powerful model offering superior performance and accuracy
• Llama 3.1 8B: Lightweight model providing excellent speed and efficiency
• Sonar 3.1 Large: Optimized for enterprise-grade applications
• Sonar 3.1 Small: Cost-effective high-performance option

All these models fully support Function Calling, enabling developers to easily build complex AI applications. Performance metrics include:

Speed Performance:

• Llama 3.1 8B leads with 158 tokens/second
• Sonar 3.1 Small follows at 145 tokens/second
• Llama 3.1 70B and Sonar 3.1 Large offer higher accuracy at slightly lower speeds

Latency Performance:

• Sonar 3.1 Large: 0.37 seconds
• Llama 3.1 8B: 0.40 seconds
• Other models maintain similarly low latency levels

Context Window Support:

• Sonar series: 131k tokens
• Llama series: 128k tokens

Pricing Strategy (per million tokens):

• Llama 3.1 8B: $0.20
• Sonar 3.1 Small: $0.20
• Llama 3.1 70B: $1.00
• Sonar 3.1 Large: $0.80

All these models support JSON mode output, making them particularly suitable for applications requiring structured data processing. By offering this diverse range of models with Function Calling support, Perplexity provides developers with flexible and powerful AI development tools.