Spaces:
Sleeping
Sleeping
Create pricing.py
Browse files- core/pricing.py +32 -0
core/pricing.py
ADDED
|
@@ -0,0 +1,32 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
from dataclasses import asdict
|
| 2 |
+
from typing import Dict, Any
|
| 3 |
+
from .config import ModelRate
|
| 4 |
+
|
| 5 |
+
def cost_from_tokens(rate: ModelRate, tokens: Dict[str, int]) -> Dict[str, Any]:
|
| 6 |
+
"""
|
| 7 |
+
tokens fields supported:
|
| 8 |
+
prompt_tokens, completion_tokens, cached_prompt_tokens, reasoning_tokens
|
| 9 |
+
"""
|
| 10 |
+
prompt = int(tokens.get("prompt_tokens", 0))
|
| 11 |
+
completion = int(tokens.get("completion_tokens", 0))
|
| 12 |
+
cached = int(tokens.get("cached_prompt_tokens", 0))
|
| 13 |
+
reasoning = int(tokens.get("reasoning_tokens", 0))
|
| 14 |
+
|
| 15 |
+
# per-token costs
|
| 16 |
+
inp = (prompt / 1_000_000.0) * rate.input_per_1m
|
| 17 |
+
out = (completion / 1_000_000.0) * rate.output_per_1m
|
| 18 |
+
cache = (cached / 1_000_000.0) * rate.cached_input_per_1m
|
| 19 |
+
reas = (reasoning / 1_000_000.0) * rate.reasoning_per_1m
|
| 20 |
+
|
| 21 |
+
total = inp + out + cache + reas
|
| 22 |
+
|
| 23 |
+
return {
|
| 24 |
+
"usd": round(total, 8),
|
| 25 |
+
"breakdown": {
|
| 26 |
+
"input": round(inp, 8),
|
| 27 |
+
"output": round(out, 8),
|
| 28 |
+
"cached_input": round(cache, 8),
|
| 29 |
+
"reasoning": round(reas, 8),
|
| 30 |
+
},
|
| 31 |
+
"rate": asdict(rate),
|
| 32 |
+
}
|