Gemini API Config w/ gemini-2.5-pro / embedding-exp-03-07

Archon
1

{
“profiles”: {
“default”: {
“LLM_PROVIDER”: “OpenAI”,
“EMBEDDING_PROVIDER”: “OpenAI”,
“BASE_URL”: “https://generativelanguage.googleapis.com/v1beta/”,
“EMBEDDING_BASE_URL”: “https://generativelanguage.googleapis.com/v1beta/”,
“SUPABASE_URL”: “XXX”,
“SUPABASE_SERVICE_KEY”: “XXX”,
“LLM_API_KEY”: “XXX”,
“PRIMARY_MODEL”: “gemini-2.5-pro-exp-03-25”,
“REASONER_MODEL”: “gemini-2.5-pro-exp-03-25”,
“EMBEDDING_API_KEY”: “XXX”,
“EMBEDDING_MODEL”: “gemini-embedding-exp-03-07”
}
},
“current_profile”: “default”
}

1 Like
2

Hey @dalkeith, could you share some context around this? :slight_smile:

3

Hey. This is a working configuration for Archon environmental variables for those who don’t have respective API’s such as OpenAI. It utilizes Gemini 2.5 pro model released yesterday, and also the new embedding model released earlier in the month. These API’s can be procured for free via Google AI studios.

2 Likes
4

Can this be added to the bolt.diy config to give access to Gemini 2.5 Pro now. It’s benchmarks are off the charts impressive…

2 Likes
5

I have been testing 2.5 it really is very very impressive, blows GPT out the water.

2 Likes
6

Agreed, super gay but impressive coding skills.

1 Like