base_model = "google/gemini-3-flash-preview"
reasoning_options = [{ type = "effort", values = ["low", "medium", "high"] }, { type = "budget_tokens", min = 256, max = 32_000 }]

[cost]
input = 0.5
output = 3
cache_read = 0.05
input_audio = 1

[limit]
context = 128_000
input = 128_000
output = 64_000