base_model = "anthropic/claude-opus-4-7"
# Chat uses `thinking = { type = "enabled", budget_tokens = N }`; N >= 1024
# and must be less than `max_tokens`.
# https://docs.databricks.com/aws/en/machine-learning/model-serving/query-reason-models (accessed 2026-06-25)
reasoning_options = [{ type = "budget_tokens", min = 1_024 }]

[cost]
input = 5
output = 25
cache_read = 0.5
cache_write = 6.25

[experimental.modes.fast]
cost = { input = 30, output = 150, cache_read = 3, cache_write = 37.5 }
provider = { body = { speed = "fast" }, headers = { anthropic-beta = "fast-mode-2026-02-01" } }