Sonar Reasoning is a high-performance reasoning model that emphasizes advanced multi-step chain-of-thought reasoning with enhanced information retrieval capabilities.
The company that provides the model
The number of tokens you can send in a prompt
The maximum number of tokens a model can generate in one request
The cost of prompt tokens sent to the model
The cost of output tokens generated by the model
When the model's knowledge ends
When the model was launched
Capability for the model to use external tools
Ability to process and analyze visual inputs, like images
Support for multiple languages
Whether the model supports fine-tuning on custom datasets
Sonar Reasoning is a high-performance reasoning model focused on advanced multi-step chain-of-thought reasoning and enhanced information retrieval capabilities.
It is best suited for analyzing complex topics that require detailed, multi-step reasoning—such as research, strategic planning, or in-depth problem solving.
Sonar Reasoning supports an 8,000-token context window, allowing for reasonably lengthy inputs and outputs.
It costs $1.00 per 1M input tokens and $5.00 per 1M output tokens.
No, it does not currently support vision capabilities or tool/function calling.
Multilingual support is not specified, and it does not support fine-tuning.
You can access the official documentation here:
https://docs.perplexity.ai/guides/models/sonar-reasoning
Collaborate with thousands of AI builders to discover, manage, and improve prompts—free to get started.
