WHich LLm used for re-phrase the user query

[bhuvneshsaini]

I entered the query "Research the latest developments in ISRO mars mission"
and agno rephrase it in 3 different queries like shown in below
Tool Calls: [
{
"id": "call_vdKMvPbNXmrNvcDcuIEJKqyt",
"type": "function",
"function": {
"name": "search_exa",
"arguments": "{"query": "ISRO Mars mission 2025",
"num_results": 5, "category": "news"}"
}
},
{
"id": "call_UjLmZlmq7YYNjbGtRXrvI8Eu",
"type": "function",
"function": {
"name": "search_exa",
"arguments": "{"query": "ISRO Mangalyaan 2 updates",
"num_results": 5, "category": "research paper"}"
}
},
{
"id": "call_hOmdlaCqUU108waN7PVrheSq",
"type": "function",
"function": {
"name": "search_exa",
"arguments": "{"query": "Indian Space Research Organisation
Mars mission plans 2025", "num_results": 5, "category": "pdf"}"
}
}
]

Will it use LLM defined by the user?
We need to calculate the intermediate hits to API to calculate the cost!!

[dirkbrnd]

Hi @bhuvneshsaini
Can you share a snippet of your code? Tool calls quality greatly depends on the model used.