jutella.toml

# OpenAI API base URL. Everything before `chat/completions` in the URL.
api_url = "https://api.openai.com/v1/"

# Auth header `Authorization: Bearer {api_token}`, used by OpenAI endpoints.
api_token = "<API token>"

# Auth header `api-key: {api_key}`, used by Azure endpoints.
# Only one of `api_key` & `api_token` must be set.
#api_key = "<API key>"

# Optional `api-version` GET parameter, used by Azure endpoints.
#api_version = "2023-03-15-preview"

# Model to use.
model = "gpt-4o-mini"

# Optional system message to initialize the model.
system_message = "You are a helpful assistant."

# Optional minimum conversation history to keep in the context.
#
# The context will be truncated to keep at least `min_history_tokens`, but
# no more than one request-response above this threshold, and under
# no circumstances more than `max_history_tokens`.
# This method of context truncation ensures that at least the latest round
# of messages is always kept (unless `max_history_tokens` kicks in).
min_history_tokens = 1000

# Optional maximum conversation history to keep in the context.
max_history_tokens = 2500

# Copy every response to clipboard via `xclip`.
xclip = false