base_model = "google/gemini-2.5-pro"
# Chat uses `thinking.budget_tokens` 128..32768; reasoning cannot be disabled;
# native Gemini -1 requests dynamic thinking but is not a numeric budget.
# https://ai.google.dev/gemini-api/docs/thinking (accessed 2026-06-25)
reasoning_options = [{ type = "budget_tokens", min = 128, max = 32_768 }]

[cost]
input = 1.25
output = 10
cache_read = 0.125

[[cost.tiers]]
tier = { type = "context", size = 200_000 }
input = 2.5
output = 15
cache_read = 0.25