Add gemini cached tokens costs for more accurate cost tracking

This commit is contained in:
Debanjum
2025-08-29 15:54:20 -07:00
parent 0babab580a
commit 3aa6f8ba1f

View File

@@ -51,8 +51,8 @@ model_to_cost: Dict[str, Dict[str, float]] = {
"gemini-2.0-flash": {"input": 0.10, "output": 0.40},
"gemini-2.0-flash-lite": {"input": 0.0075, "output": 0.30},
"gemini-2.5-flash-lite": {"input": 0.10, "output": 0.40},
"gemini-2.5-flash": {"input": 0.30, "output": 2.50},
"gemini-2.5-pro": {"input": 1.25, "output": 10.0},
"gemini-2.5-flash": {"input": 0.30, "cache_read_tokens": 0.075, "output": 2.50},
"gemini-2.5-pro": {"input": 1.25, "cache_read_tokens": 0.31, "output": 10.0},
# Anthropic Pricing: https://www.anthropic.com/pricing#anthropic-api
"claude-3-5-haiku-20241022": {"input": 1.0, "output": 5.0, "cache_read": 0.08, "cache_write": 1.0},
"claude-3-5-haiku@20241022": {"input": 1.0, "output": 5.0, "cache_read": 0.08, "cache_write": 1.0},