Coverage for little_loops / pricing.py: 38%

1"""Model pricing constants for token cost estimation. 

2 

3Prices are in USD per million tokens ($/Mtok). 

4Source: Anthropic pricing page (as of June 2026). 

5""" 

6 

7from __future__ import annotations 

8 

9# Per-model pricing: {model_id: {token_type: usd_per_million}} 

10MODEL_PRICING: dict[str, dict[str, float]] = { 

11 # Claude 4.x 

12 "claude-opus-4-7": { 

13 "input": 15.0, 

14 "output": 75.0, 

15 "cache_read": 1.50, 

16 "cache_creation": 18.75, 

17 }, 

18 "claude-opus-4-6": { 

19 "input": 15.0, 

20 "output": 75.0, 

21 "cache_read": 1.50, 

22 "cache_creation": 18.75, 

23 }, 

24 "claude-sonnet-4-6": { 

25 "input": 3.0, 

26 "output": 15.0, 

27 "cache_read": 0.30, 

28 "cache_creation": 3.75, 

29 }, 

30 "claude-haiku-4-5-20251001": { 

31 "input": 0.80, 

32 "output": 4.0, 

33 "cache_read": 0.08, 

34 "cache_creation": 1.0, 

35 }, 

36 # Claude 3.x (legacy, may still appear in logs) 

37 "claude-opus-4-5": { 

38 "input": 15.0, 

39 "output": 75.0, 

40 "cache_read": 1.50, 

41 "cache_creation": 18.75, 

42 }, 

43 "claude-sonnet-3-7": { 

44 "input": 3.0, 

45 "output": 15.0, 

46 "cache_read": 0.30, 

47 "cache_creation": 3.75, 

48 }, 

49 "claude-haiku-3-5": { 

50 "input": 0.80, 

51 "output": 4.0, 

52 "cache_read": 0.08, 

53 "cache_creation": 1.0, 

54 }, 

55} 

56 

57 

58def estimate_cost_usd( 

59 model: str, 

60 input_tokens: int, 

61 output_tokens: int, 

62 cache_read_tokens: int = 0, 

63 cache_creation_tokens: int = 0, 

64) -> float | None: 

65 """Estimate cost in USD for a token usage event. 

66 

67 Returns None if the model is not in MODEL_PRICING. 

68 """ 

69 pricing = MODEL_PRICING.get(model) 

70 if pricing is None: 

71 return None 

72 per_m = 1_000_000.0 

73 return ( 

74 input_tokens * pricing["input"] / per_m 

75 + output_tokens * pricing["output"] / per_m 

76 + cache_read_tokens * pricing["cache_read"] / per_m 

77 + cache_creation_tokens * pricing["cache_creation"] / per_m 

78 )