-
Notifications
You must be signed in to change notification settings - Fork 803
Open
Labels
Description
Describe the feature you want to propose
Currently, models and providers are bounded. I am thinking to support the following configuration format:
trae-agent:
max_steps: 20
enable_lakeview: true
model: model_1
another-feature-agent:
max_steps: 20
enable_lakeview: true
model: model_1
model_providers:
openai:
api_key: your_openai_api_key
base_url: base_url
anthropic:
api_key: your_anthropic_api_key
base_url: base_url
models:
model_1:
model_name: gpt-4o
provider: openai
max_tokens: 128000
The benefit is that we can also configure agents, with flexible model selections.
What problem does this feature solve or what benefit does it bring?
Flexible agent and model configuration
Implementation details or suggestions (optional)
No response
yihong0618, rupaut98 and luke396