llm_cost_tracker 0.4.1 → 0.5.1
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- checksums.yaml +4 -4
- data/CHANGELOG.md +30 -0
- data/README.md +132 -405
- data/lib/llm_cost_tracker/configuration/instrumentation.rb +37 -0
- data/lib/llm_cost_tracker/configuration.rb +10 -5
- data/lib/llm_cost_tracker/doctor.rb +166 -0
- data/lib/llm_cost_tracker/generators/llm_cost_tracker/install_generator.rb +33 -0
- data/lib/llm_cost_tracker/generators/llm_cost_tracker/prices_generator.rb +12 -6
- data/lib/llm_cost_tracker/generators/llm_cost_tracker/templates/initializer.rb.erb +53 -21
- data/lib/llm_cost_tracker/integrations/anthropic.rb +75 -0
- data/lib/llm_cost_tracker/integrations/base.rb +72 -0
- data/lib/llm_cost_tracker/integrations/object_reader.rb +56 -0
- data/lib/llm_cost_tracker/integrations/openai.rb +95 -0
- data/lib/llm_cost_tracker/integrations/registry.rb +41 -0
- data/lib/llm_cost_tracker/middleware/faraday.rb +6 -5
- data/lib/llm_cost_tracker/parsed_usage.rb +8 -1
- data/lib/llm_cost_tracker/parsers/base.rb +1 -1
- data/lib/llm_cost_tracker/parsers/openai_usage.rb +1 -1
- data/lib/llm_cost_tracker/price_freshness.rb +38 -0
- data/lib/llm_cost_tracker/price_registry.rb +14 -0
- data/lib/llm_cost_tracker/price_sync/fetcher.rb +5 -2
- data/lib/llm_cost_tracker/price_sync/registry_diff.rb +51 -0
- data/lib/llm_cost_tracker/price_sync/registry_writer.rb +5 -1
- data/lib/llm_cost_tracker/price_sync.rb +111 -109
- data/lib/llm_cost_tracker/prices.json +391 -42
- data/lib/llm_cost_tracker/pricing.rb +35 -16
- data/lib/llm_cost_tracker/request_url.rb +20 -0
- data/lib/llm_cost_tracker/storage/dispatcher.rb +68 -0
- data/lib/llm_cost_tracker/stream_collector.rb +3 -3
- data/lib/llm_cost_tracker/tag_context.rb +52 -0
- data/lib/llm_cost_tracker/tracker.rb +7 -60
- data/lib/llm_cost_tracker/version.rb +1 -1
- data/lib/llm_cost_tracker.rb +14 -4
- data/lib/tasks/llm_cost_tracker.rake +33 -69
- metadata +28 -12
- data/lib/llm_cost_tracker/generators/llm_cost_tracker/templates/llm_cost_tracker_prices.yml.erb +0 -51
- data/lib/llm_cost_tracker/price_sync/merger.rb +0 -72
- data/lib/llm_cost_tracker/price_sync/model_catalog.rb +0 -77
- data/lib/llm_cost_tracker/price_sync/raw_price.rb +0 -33
- data/lib/llm_cost_tracker/price_sync/refresh_plan_builder.rb +0 -162
- data/lib/llm_cost_tracker/price_sync/source.rb +0 -29
- data/lib/llm_cost_tracker/price_sync/source_result.rb +0 -7
- data/lib/llm_cost_tracker/price_sync/sources/litellm.rb +0 -90
- data/lib/llm_cost_tracker/price_sync/sources/open_router.rb +0 -93
- data/lib/llm_cost_tracker/price_sync/validator.rb +0 -66
|
@@ -1,51 +1,400 @@
|
|
|
1
1
|
{
|
|
2
2
|
"metadata": {
|
|
3
|
-
"updated_at": "2026-04-
|
|
3
|
+
"updated_at": "2026-04-27",
|
|
4
4
|
"currency": "USD",
|
|
5
5
|
"unit": "1M tokens",
|
|
6
6
|
"source_urls": [
|
|
7
|
-
"https://openai.com/api/pricing",
|
|
8
|
-
"https://
|
|
9
|
-
"https://ai.google.dev/
|
|
10
|
-
]
|
|
7
|
+
"https://developers.openai.com/api/docs/pricing",
|
|
8
|
+
"https://platform.claude.com/docs/en/about-claude/pricing",
|
|
9
|
+
"https://ai.google.dev/pricing"
|
|
10
|
+
],
|
|
11
|
+
"schema_version": 1,
|
|
12
|
+
"min_gem_version": "0.4.0"
|
|
11
13
|
},
|
|
12
14
|
"models": {
|
|
13
|
-
"
|
|
14
|
-
|
|
15
|
-
|
|
16
|
-
|
|
17
|
-
|
|
18
|
-
|
|
19
|
-
|
|
20
|
-
|
|
21
|
-
"
|
|
22
|
-
|
|
23
|
-
|
|
24
|
-
|
|
25
|
-
|
|
26
|
-
|
|
27
|
-
|
|
28
|
-
|
|
29
|
-
"
|
|
30
|
-
|
|
31
|
-
|
|
32
|
-
|
|
33
|
-
|
|
34
|
-
|
|
35
|
-
|
|
36
|
-
|
|
37
|
-
"claude-
|
|
38
|
-
|
|
39
|
-
|
|
40
|
-
|
|
41
|
-
|
|
42
|
-
|
|
43
|
-
|
|
44
|
-
|
|
45
|
-
"
|
|
46
|
-
|
|
47
|
-
|
|
48
|
-
|
|
49
|
-
|
|
15
|
+
"anthropic/claude-haiku-4-5": {
|
|
16
|
+
"input": 1.0,
|
|
17
|
+
"output": 5.0,
|
|
18
|
+
"cache_read_input": 0.1,
|
|
19
|
+
"cache_write_input": 1.25,
|
|
20
|
+
"batch_input": 0.5,
|
|
21
|
+
"batch_output": 2.5
|
|
22
|
+
},
|
|
23
|
+
"anthropic/claude-opus-4": {
|
|
24
|
+
"input": 15.0,
|
|
25
|
+
"output": 75.0,
|
|
26
|
+
"cache_read_input": 1.5,
|
|
27
|
+
"cache_write_input": 18.75,
|
|
28
|
+
"batch_input": 7.5,
|
|
29
|
+
"batch_output": 37.5
|
|
30
|
+
},
|
|
31
|
+
"anthropic/claude-opus-4-1": {
|
|
32
|
+
"input": 15.0,
|
|
33
|
+
"output": 75.0,
|
|
34
|
+
"cache_read_input": 1.5,
|
|
35
|
+
"cache_write_input": 18.75,
|
|
36
|
+
"batch_input": 7.5,
|
|
37
|
+
"batch_output": 37.5
|
|
38
|
+
},
|
|
39
|
+
"anthropic/claude-opus-4-5": {
|
|
40
|
+
"input": 5.0,
|
|
41
|
+
"output": 25.0,
|
|
42
|
+
"cache_read_input": 0.5,
|
|
43
|
+
"cache_write_input": 6.25,
|
|
44
|
+
"batch_input": 2.5,
|
|
45
|
+
"batch_output": 12.5
|
|
46
|
+
},
|
|
47
|
+
"anthropic/claude-opus-4-6": {
|
|
48
|
+
"input": 5.0,
|
|
49
|
+
"output": 25.0,
|
|
50
|
+
"cache_read_input": 0.5,
|
|
51
|
+
"cache_write_input": 6.25,
|
|
52
|
+
"batch_input": 2.5,
|
|
53
|
+
"batch_output": 12.5
|
|
54
|
+
},
|
|
55
|
+
"anthropic/claude-opus-4-7": {
|
|
56
|
+
"input": 5.0,
|
|
57
|
+
"output": 25.0,
|
|
58
|
+
"cache_read_input": 0.5,
|
|
59
|
+
"cache_write_input": 6.25,
|
|
60
|
+
"batch_input": 2.5,
|
|
61
|
+
"batch_output": 12.5
|
|
62
|
+
},
|
|
63
|
+
"anthropic/claude-sonnet-4": {
|
|
64
|
+
"input": 3.0,
|
|
65
|
+
"output": 15.0,
|
|
66
|
+
"cache_read_input": 0.3,
|
|
67
|
+
"cache_write_input": 3.75,
|
|
68
|
+
"batch_input": 1.5,
|
|
69
|
+
"batch_output": 7.5
|
|
70
|
+
},
|
|
71
|
+
"anthropic/claude-sonnet-4-5": {
|
|
72
|
+
"input": 3.0,
|
|
73
|
+
"output": 15.0,
|
|
74
|
+
"cache_read_input": 0.3,
|
|
75
|
+
"cache_write_input": 3.75,
|
|
76
|
+
"batch_input": 1.5,
|
|
77
|
+
"batch_output": 7.5
|
|
78
|
+
},
|
|
79
|
+
"anthropic/claude-sonnet-4-6": {
|
|
80
|
+
"input": 3.0,
|
|
81
|
+
"output": 15.0,
|
|
82
|
+
"cache_read_input": 0.3,
|
|
83
|
+
"cache_write_input": 3.75,
|
|
84
|
+
"batch_input": 1.5,
|
|
85
|
+
"batch_output": 7.5
|
|
86
|
+
},
|
|
87
|
+
"gemini/gemini-2.0-flash": {
|
|
88
|
+
"input": 0.1,
|
|
89
|
+
"cache_read_input": 0.025,
|
|
90
|
+
"output": 0.4,
|
|
91
|
+
"batch_input": 0.05,
|
|
92
|
+
"batch_output": 0.2
|
|
93
|
+
},
|
|
94
|
+
"gemini/gemini-2.0-flash-lite": {
|
|
95
|
+
"input": 0.075,
|
|
96
|
+
"output": 0.3,
|
|
97
|
+
"batch_input": 0.0375,
|
|
98
|
+
"batch_output": 0.15
|
|
99
|
+
},
|
|
100
|
+
"gemini/gemini-2.5-flash": {
|
|
101
|
+
"input": 0.3,
|
|
102
|
+
"output": 2.5,
|
|
103
|
+
"cache_read_input": 0.03,
|
|
104
|
+
"cache_write_input": 0.083333333333,
|
|
105
|
+
"batch_input": 0.15,
|
|
106
|
+
"batch_output": 1.25
|
|
107
|
+
},
|
|
108
|
+
"gemini/gemini-2.5-flash-lite": {
|
|
109
|
+
"input": 0.1,
|
|
110
|
+
"output": 0.4,
|
|
111
|
+
"cache_read_input": 0.01,
|
|
112
|
+
"cache_write_input": 0.083333333333,
|
|
113
|
+
"batch_input": 0.05,
|
|
114
|
+
"batch_output": 0.2
|
|
115
|
+
},
|
|
116
|
+
"gemini/gemini-2.5-pro": {
|
|
117
|
+
"input": 1.25,
|
|
118
|
+
"output": 10.0,
|
|
119
|
+
"cache_read_input": 0.125,
|
|
120
|
+
"cache_write_input": 0.375,
|
|
121
|
+
"batch_input": 0.625,
|
|
122
|
+
"batch_output": 5.0
|
|
123
|
+
},
|
|
124
|
+
"openai/gpt-3.5-turbo": {
|
|
125
|
+
"input": 0.5,
|
|
126
|
+
"output": 1.5
|
|
127
|
+
},
|
|
128
|
+
"openai/gpt-4": {
|
|
129
|
+
"input": 30.0,
|
|
130
|
+
"output": 60.0,
|
|
131
|
+
"batch_input": 15.0,
|
|
132
|
+
"batch_output": 30.0
|
|
133
|
+
},
|
|
134
|
+
"openai/gpt-4-turbo": {
|
|
135
|
+
"input": 10.0,
|
|
136
|
+
"output": 30.0,
|
|
137
|
+
"batch_input": 5.0,
|
|
138
|
+
"batch_output": 15.0
|
|
139
|
+
},
|
|
140
|
+
"openai/gpt-4.1": {
|
|
141
|
+
"input": 2.0,
|
|
142
|
+
"output": 8.0,
|
|
143
|
+
"cache_read_input": 0.5,
|
|
144
|
+
"batch_input": 1.0,
|
|
145
|
+
"batch_output": 4.0
|
|
146
|
+
},
|
|
147
|
+
"openai/gpt-4.1-mini": {
|
|
148
|
+
"input": 0.4,
|
|
149
|
+
"output": 1.6,
|
|
150
|
+
"cache_read_input": 0.1,
|
|
151
|
+
"batch_input": 0.2,
|
|
152
|
+
"batch_output": 0.8
|
|
153
|
+
},
|
|
154
|
+
"openai/gpt-4.1-nano": {
|
|
155
|
+
"input": 0.1,
|
|
156
|
+
"output": 0.4,
|
|
157
|
+
"cache_read_input": 0.025,
|
|
158
|
+
"batch_input": 0.05,
|
|
159
|
+
"batch_output": 0.2
|
|
160
|
+
},
|
|
161
|
+
"openai/gpt-4o": {
|
|
162
|
+
"input": 2.5,
|
|
163
|
+
"output": 10.0,
|
|
164
|
+
"cache_read_input": 1.25,
|
|
165
|
+
"batch_input": 1.25,
|
|
166
|
+
"batch_output": 5.0
|
|
167
|
+
},
|
|
168
|
+
"openai/gpt-4o-2024-05-13": {
|
|
169
|
+
"input": 5.0,
|
|
170
|
+
"output": 15.0,
|
|
171
|
+
"batch_input": 2.5,
|
|
172
|
+
"batch_output": 7.5
|
|
173
|
+
},
|
|
174
|
+
"openai/gpt-4o-mini": {
|
|
175
|
+
"input": 0.15,
|
|
176
|
+
"output": 0.6,
|
|
177
|
+
"cache_read_input": 0.075,
|
|
178
|
+
"batch_input": 0.075,
|
|
179
|
+
"batch_output": 0.3
|
|
180
|
+
},
|
|
181
|
+
"openai/gpt-5": {
|
|
182
|
+
"input": 1.25,
|
|
183
|
+
"output": 10.0,
|
|
184
|
+
"cache_read_input": 0.125,
|
|
185
|
+
"batch_input": 0.625,
|
|
186
|
+
"batch_output": 5.0,
|
|
187
|
+
"batch_cache_read_input": 0.0625
|
|
188
|
+
},
|
|
189
|
+
"openai/gpt-5-chat-latest": {
|
|
190
|
+
"input": 1.25,
|
|
191
|
+
"output": 10.0,
|
|
192
|
+
"cache_read_input": 0.125
|
|
193
|
+
},
|
|
194
|
+
"openai/gpt-5-codex": {
|
|
195
|
+
"input": 1.25,
|
|
196
|
+
"output": 10.0,
|
|
197
|
+
"cache_read_input": 0.125
|
|
198
|
+
},
|
|
199
|
+
"openai/gpt-5-mini": {
|
|
200
|
+
"input": 0.25,
|
|
201
|
+
"output": 2.0,
|
|
202
|
+
"cache_read_input": 0.025,
|
|
203
|
+
"batch_input": 0.125,
|
|
204
|
+
"batch_output": 1.0,
|
|
205
|
+
"batch_cache_read_input": 0.0125
|
|
206
|
+
},
|
|
207
|
+
"openai/gpt-5-nano": {
|
|
208
|
+
"input": 0.05,
|
|
209
|
+
"output": 0.4,
|
|
210
|
+
"cache_read_input": 0.005,
|
|
211
|
+
"batch_input": 0.025,
|
|
212
|
+
"batch_output": 0.2,
|
|
213
|
+
"batch_cache_read_input": 0.0025
|
|
214
|
+
},
|
|
215
|
+
"openai/gpt-5-pro": {
|
|
216
|
+
"input": 15.0,
|
|
217
|
+
"output": 120.0,
|
|
218
|
+
"batch_input": 7.5,
|
|
219
|
+
"batch_output": 60.0
|
|
220
|
+
},
|
|
221
|
+
"openai/gpt-5.1": {
|
|
222
|
+
"input": 1.25,
|
|
223
|
+
"output": 10.0,
|
|
224
|
+
"cache_read_input": 0.125,
|
|
225
|
+
"batch_input": 0.625,
|
|
226
|
+
"batch_output": 5.0,
|
|
227
|
+
"batch_cache_read_input": 0.0625
|
|
228
|
+
},
|
|
229
|
+
"openai/gpt-5.1-chat-latest": {
|
|
230
|
+
"input": 1.25,
|
|
231
|
+
"output": 10.0,
|
|
232
|
+
"cache_read_input": 0.125
|
|
233
|
+
},
|
|
234
|
+
"openai/gpt-5.1-codex": {
|
|
235
|
+
"input": 1.25,
|
|
236
|
+
"output": 10.0,
|
|
237
|
+
"cache_read_input": 0.125
|
|
238
|
+
},
|
|
239
|
+
"openai/gpt-5.1-codex-max": {
|
|
240
|
+
"input": 1.25,
|
|
241
|
+
"output": 10.0,
|
|
242
|
+
"cache_read_input": 0.125
|
|
243
|
+
},
|
|
244
|
+
"openai/gpt-5.1-codex-mini": {
|
|
245
|
+
"input": 0.25,
|
|
246
|
+
"output": 2.0,
|
|
247
|
+
"cache_read_input": 0.025
|
|
248
|
+
},
|
|
249
|
+
"openai/gpt-5.2": {
|
|
250
|
+
"input": 1.75,
|
|
251
|
+
"output": 14.0,
|
|
252
|
+
"cache_read_input": 0.175,
|
|
253
|
+
"batch_input": 0.875,
|
|
254
|
+
"batch_output": 7.0,
|
|
255
|
+
"batch_cache_read_input": 0.0875
|
|
256
|
+
},
|
|
257
|
+
"openai/gpt-5.2-chat-latest": {
|
|
258
|
+
"input": 1.75,
|
|
259
|
+
"output": 14.0,
|
|
260
|
+
"cache_read_input": 0.175
|
|
261
|
+
},
|
|
262
|
+
"openai/gpt-5.2-codex": {
|
|
263
|
+
"input": 1.75,
|
|
264
|
+
"output": 14.0,
|
|
265
|
+
"cache_read_input": 0.175
|
|
266
|
+
},
|
|
267
|
+
"openai/gpt-5.2-pro": {
|
|
268
|
+
"input": 21.0,
|
|
269
|
+
"output": 168.0,
|
|
270
|
+
"batch_input": 10.5,
|
|
271
|
+
"batch_output": 84.0
|
|
272
|
+
},
|
|
273
|
+
"openai/gpt-5.4": {
|
|
274
|
+
"input": 2.5,
|
|
275
|
+
"output": 15.0,
|
|
276
|
+
"cache_read_input": 0.25,
|
|
277
|
+
"batch_input": 1.25,
|
|
278
|
+
"batch_output": 7.5,
|
|
279
|
+
"batch_cache_read_input": 0.13
|
|
280
|
+
},
|
|
281
|
+
"openai/gpt-5.4-mini": {
|
|
282
|
+
"input": 0.75,
|
|
283
|
+
"output": 4.5,
|
|
284
|
+
"cache_read_input": 0.075,
|
|
285
|
+
"batch_input": 0.375,
|
|
286
|
+
"batch_output": 2.25,
|
|
287
|
+
"batch_cache_read_input": 0.0375
|
|
288
|
+
},
|
|
289
|
+
"openai/gpt-5.4-nano": {
|
|
290
|
+
"input": 0.2,
|
|
291
|
+
"output": 1.25,
|
|
292
|
+
"cache_read_input": 0.02,
|
|
293
|
+
"batch_input": 0.1,
|
|
294
|
+
"batch_output": 0.625,
|
|
295
|
+
"batch_cache_read_input": 0.01
|
|
296
|
+
},
|
|
297
|
+
"openai/gpt-5.4-pro": {
|
|
298
|
+
"input": 30.0,
|
|
299
|
+
"output": 180.0,
|
|
300
|
+
"batch_input": 15.0,
|
|
301
|
+
"batch_output": 90.0
|
|
302
|
+
},
|
|
303
|
+
"openai/gpt-5.5": {
|
|
304
|
+
"input": 5.0,
|
|
305
|
+
"output": 30.0,
|
|
306
|
+
"cache_read_input": 0.5,
|
|
307
|
+
"batch_input": 2.5,
|
|
308
|
+
"batch_output": 15.0,
|
|
309
|
+
"batch_cache_read_input": 0.25
|
|
310
|
+
},
|
|
311
|
+
"openai/gpt-5.5-pro": {
|
|
312
|
+
"input": 30.0,
|
|
313
|
+
"output": 180.0,
|
|
314
|
+
"batch_input": 15.0,
|
|
315
|
+
"batch_output": 90.0
|
|
316
|
+
},
|
|
317
|
+
"openai/o1": {
|
|
318
|
+
"input": 15.0,
|
|
319
|
+
"output": 60.0,
|
|
320
|
+
"cache_read_input": 7.5,
|
|
321
|
+
"batch_input": 7.5,
|
|
322
|
+
"batch_output": 30.0
|
|
323
|
+
},
|
|
324
|
+
"openai/o1-mini": {
|
|
325
|
+
"input": 1.1,
|
|
326
|
+
"cache_read_input": 0.55,
|
|
327
|
+
"output": 4.4,
|
|
328
|
+
"batch_input": 0.55,
|
|
329
|
+
"batch_output": 2.2
|
|
330
|
+
},
|
|
331
|
+
"openai/o3": {
|
|
332
|
+
"input": 2.0,
|
|
333
|
+
"output": 8.0,
|
|
334
|
+
"cache_read_input": 0.5,
|
|
335
|
+
"batch_input": 1.0,
|
|
336
|
+
"batch_output": 4.0
|
|
337
|
+
},
|
|
338
|
+
"openai/o3-mini": {
|
|
339
|
+
"input": 1.1,
|
|
340
|
+
"output": 4.4,
|
|
341
|
+
"cache_read_input": 0.55,
|
|
342
|
+
"batch_input": 0.55,
|
|
343
|
+
"batch_output": 2.2
|
|
344
|
+
},
|
|
345
|
+
"openai/o4-mini": {
|
|
346
|
+
"input": 1.1,
|
|
347
|
+
"output": 4.4,
|
|
348
|
+
"cache_read_input": 0.275,
|
|
349
|
+
"batch_input": 0.55,
|
|
350
|
+
"batch_output": 2.2
|
|
351
|
+
},
|
|
352
|
+
"anthropic/claude-haiku-3-5": {
|
|
353
|
+
"input": 0.8,
|
|
354
|
+
"cache_write_input": 1.0,
|
|
355
|
+
"cache_read_input": 0.08,
|
|
356
|
+
"output": 4.0,
|
|
357
|
+
"batch_input": 0.4,
|
|
358
|
+
"batch_output": 2.0
|
|
359
|
+
},
|
|
360
|
+
"anthropic/claude-haiku-3": {
|
|
361
|
+
"input": 0.25,
|
|
362
|
+
"cache_write_input": 0.3,
|
|
363
|
+
"cache_read_input": 0.03,
|
|
364
|
+
"output": 1.25,
|
|
365
|
+
"batch_input": 0.125,
|
|
366
|
+
"batch_output": 0.625
|
|
367
|
+
},
|
|
368
|
+
"openai/o1-pro": {
|
|
369
|
+
"input": 150.0,
|
|
370
|
+
"output": 600.0,
|
|
371
|
+
"batch_input": 75.0,
|
|
372
|
+
"batch_output": 300.0
|
|
373
|
+
},
|
|
374
|
+
"openai/o3-pro": {
|
|
375
|
+
"input": 20.0,
|
|
376
|
+
"output": 80.0,
|
|
377
|
+
"batch_input": 10.0,
|
|
378
|
+
"batch_output": 40.0
|
|
379
|
+
},
|
|
380
|
+
"openai/gpt-5.3-chat-latest": {
|
|
381
|
+
"input": 1.75,
|
|
382
|
+
"output": 14.0,
|
|
383
|
+
"cache_read_input": 0.175
|
|
384
|
+
},
|
|
385
|
+
"openai/chatgpt-4o-latest": {
|
|
386
|
+
"input": 5.0,
|
|
387
|
+
"output": 15.0
|
|
388
|
+
},
|
|
389
|
+
"openai/gpt-5.3-codex": {
|
|
390
|
+
"input": 1.75,
|
|
391
|
+
"output": 14.0,
|
|
392
|
+
"cache_read_input": 0.175
|
|
393
|
+
},
|
|
394
|
+
"openai/codex-mini-latest": {
|
|
395
|
+
"input": 1.5,
|
|
396
|
+
"output": 6.0,
|
|
397
|
+
"cache_read_input": 0.375
|
|
398
|
+
}
|
|
50
399
|
}
|
|
51
400
|
}
|
|
@@ -32,27 +32,20 @@ module LlmCostTracker
|
|
|
32
32
|
end
|
|
33
33
|
|
|
34
34
|
def lookup(provider:, model:)
|
|
35
|
-
table = prices
|
|
36
35
|
provider_name = provider.to_s
|
|
37
36
|
model_name = model.to_s
|
|
38
37
|
provider_model = provider_name.empty? ? model_name : "#{provider_name}/#{model_name}"
|
|
39
38
|
normalized_model = normalize_model_name(model_name)
|
|
39
|
+
current = current_price_tables
|
|
40
40
|
|
|
41
|
-
|
|
42
|
-
|
|
43
|
-
|
|
44
|
-
fuzzy_match(provider_model, normalized_model, table)
|
|
45
|
-
end
|
|
46
|
-
|
|
47
|
-
def models
|
|
48
|
-
prices.keys
|
|
41
|
+
lookup_in_table(current.fetch(:pricing_overrides), provider_model, model_name, normalized_model) ||
|
|
42
|
+
lookup_in_table(current.fetch(:file_prices), provider_model, model_name, normalized_model) ||
|
|
43
|
+
lookup_in_table(PRICES, provider_model, model_name, normalized_model)
|
|
49
44
|
end
|
|
50
45
|
|
|
51
|
-
|
|
52
|
-
PriceRegistry.metadata
|
|
53
|
-
end
|
|
46
|
+
private
|
|
54
47
|
|
|
55
|
-
def
|
|
48
|
+
def current_price_tables
|
|
56
49
|
file_prices = PriceRegistry.file_prices(LlmCostTracker.configuration.prices_file)
|
|
57
50
|
overrides = PriceRegistry.normalize_price_table(LlmCostTracker.configuration.pricing_overrides)
|
|
58
51
|
cache_key = [file_prices.object_id, LlmCostTracker.configuration.pricing_overrides.hash]
|
|
@@ -64,13 +57,22 @@ module LlmCostTracker
|
|
|
64
57
|
cached = @prices_cache
|
|
65
58
|
return cached[:value] if cached && cached[:key] == cache_key
|
|
66
59
|
|
|
67
|
-
value =
|
|
60
|
+
value = { pricing_overrides: overrides, file_prices: file_prices }.freeze
|
|
68
61
|
@prices_cache = { key: cache_key, value: value }.freeze
|
|
69
62
|
value
|
|
70
63
|
end
|
|
71
64
|
end
|
|
72
65
|
|
|
73
|
-
|
|
66
|
+
def lookup_in_table(table, provider_model, model_name, normalized_model)
|
|
67
|
+
return nil if table.empty?
|
|
68
|
+
|
|
69
|
+
table[provider_model] ||
|
|
70
|
+
table[model_name] ||
|
|
71
|
+
table[normalized_model] ||
|
|
72
|
+
unique_providerless_lookup(normalized_model, table) ||
|
|
73
|
+
fuzzy_match(provider_model, normalized_model, table) ||
|
|
74
|
+
unique_providerless_fuzzy_match(normalized_model, table)
|
|
75
|
+
end
|
|
74
76
|
|
|
75
77
|
def calculate_costs(usage, prices, pricing_mode:)
|
|
76
78
|
{
|
|
@@ -113,14 +115,31 @@ module LlmCostTracker
|
|
|
113
115
|
model.to_s.split("/").last
|
|
114
116
|
end
|
|
115
117
|
|
|
118
|
+
def unique_providerless_lookup(model, table)
|
|
119
|
+
matches = sorted_price_keys(table).select { |key| normalize_model_name(key) == model }
|
|
120
|
+
table[matches.first] if matches.one?
|
|
121
|
+
end
|
|
122
|
+
|
|
116
123
|
def fuzzy_match(model, normalized_model, table)
|
|
117
124
|
sorted_price_keys(table).each do |key|
|
|
118
|
-
return table[key] if
|
|
125
|
+
return table[key] if snapshot_variant?(model, key) || snapshot_variant?(normalized_model, key)
|
|
119
126
|
end
|
|
120
127
|
|
|
121
128
|
nil
|
|
122
129
|
end
|
|
123
130
|
|
|
131
|
+
def unique_providerless_fuzzy_match(model, table)
|
|
132
|
+
matches = sorted_price_keys(table).select { |key| snapshot_variant?(model, normalize_model_name(key)) }
|
|
133
|
+
table[matches.first] if matches.one?
|
|
134
|
+
end
|
|
135
|
+
|
|
136
|
+
def snapshot_variant?(model, key)
|
|
137
|
+
suffix = model.delete_prefix("#{key}-")
|
|
138
|
+
return false if suffix == model
|
|
139
|
+
|
|
140
|
+
suffix.match?(/\A(?:\d{4}-\d{2}-\d{2}|\d{8})\z/)
|
|
141
|
+
end
|
|
142
|
+
|
|
124
143
|
def sorted_price_keys(table)
|
|
125
144
|
cached = @sorted_price_keys_cache
|
|
126
145
|
return cached[:keys] if cached && cached[:table].equal?(table)
|
|
@@ -0,0 +1,20 @@
|
|
|
1
|
+
# frozen_string_literal: true
|
|
2
|
+
|
|
3
|
+
require "uri"
|
|
4
|
+
|
|
5
|
+
module LlmCostTracker
|
|
6
|
+
module RequestUrl
|
|
7
|
+
class << self
|
|
8
|
+
def label(value)
|
|
9
|
+
uri = URI.parse(value.to_s)
|
|
10
|
+
uri.query = nil
|
|
11
|
+
uri.fragment = nil
|
|
12
|
+
uri.user = nil if uri.respond_to?(:user=)
|
|
13
|
+
uri.password = nil if uri.respond_to?(:password=)
|
|
14
|
+
uri.to_s
|
|
15
|
+
rescue URI::InvalidURIError
|
|
16
|
+
value.to_s.split("?", 2).first
|
|
17
|
+
end
|
|
18
|
+
end
|
|
19
|
+
end
|
|
20
|
+
end
|
|
@@ -0,0 +1,68 @@
|
|
|
1
|
+
# frozen_string_literal: true
|
|
2
|
+
|
|
3
|
+
require_relative "../logging"
|
|
4
|
+
|
|
5
|
+
module LlmCostTracker
|
|
6
|
+
module Storage
|
|
7
|
+
class Dispatcher
|
|
8
|
+
class << self
|
|
9
|
+
def save(event)
|
|
10
|
+
config = LlmCostTracker.configuration
|
|
11
|
+
case config.storage_backend
|
|
12
|
+
when :log then log_event(event, config)
|
|
13
|
+
when :active_record then active_record_save(event)
|
|
14
|
+
when :custom then custom_save(event, config)
|
|
15
|
+
end
|
|
16
|
+
rescue LlmCostTracker::BudgetExceededError, LlmCostTracker::UnknownPricingError
|
|
17
|
+
raise
|
|
18
|
+
rescue StandardError => e
|
|
19
|
+
handle_error(e)
|
|
20
|
+
false
|
|
21
|
+
end
|
|
22
|
+
|
|
23
|
+
private
|
|
24
|
+
|
|
25
|
+
def log_event(event, config)
|
|
26
|
+
message = "#{event.provider}/#{event.model} " \
|
|
27
|
+
"tokens=#{event.total_tokens} " \
|
|
28
|
+
"cost=#{log_cost_label(event)}"
|
|
29
|
+
message += " latency=#{event.latency_ms}ms" if event.latency_ms
|
|
30
|
+
message += " stream=#{event.stream}" if event.stream
|
|
31
|
+
message += " source=#{event.usage_source}" if event.usage_source
|
|
32
|
+
message += " tags=#{event.tags}" unless event.tags.empty?
|
|
33
|
+
|
|
34
|
+
Logging.log(config.log_level, message)
|
|
35
|
+
event
|
|
36
|
+
end
|
|
37
|
+
|
|
38
|
+
def log_cost_label(event) = event.cost ? "$#{format('%.6f', event.cost.total_cost)}" : "unknown"
|
|
39
|
+
|
|
40
|
+
def active_record_save(event)
|
|
41
|
+
require_relative "../llm_api_call" unless defined?(LlmCostTracker::LlmApiCall)
|
|
42
|
+
require_relative "active_record_store" unless defined?(LlmCostTracker::Storage::ActiveRecordStore)
|
|
43
|
+
|
|
44
|
+
ActiveRecordStore.save(event)
|
|
45
|
+
event
|
|
46
|
+
rescue LoadError => e
|
|
47
|
+
raise Error, "ActiveRecord storage requires the active_record gem: #{e.message}"
|
|
48
|
+
end
|
|
49
|
+
|
|
50
|
+
def custom_save(event, config)
|
|
51
|
+
result = config.custom_storage&.call(event)
|
|
52
|
+
result == false ? false : event
|
|
53
|
+
end
|
|
54
|
+
|
|
55
|
+
def handle_error(error)
|
|
56
|
+
case LlmCostTracker.configuration.storage_error_behavior
|
|
57
|
+
when :ignore
|
|
58
|
+
nil
|
|
59
|
+
when :warn
|
|
60
|
+
Logging.warn("Storage failed; tracking event was not persisted: #{error.class}: #{error.message}")
|
|
61
|
+
when :raise
|
|
62
|
+
raise StorageError, error
|
|
63
|
+
end
|
|
64
|
+
end
|
|
65
|
+
end
|
|
66
|
+
end
|
|
67
|
+
end
|
|
68
|
+
end
|
|
@@ -115,7 +115,7 @@ module LlmCostTracker
|
|
|
115
115
|
def finalize(parsed, snapshot)
|
|
116
116
|
parsed.with(
|
|
117
117
|
provider: @provider,
|
|
118
|
-
model: present_model(parsed.model) || snapshot[:model]
|
|
118
|
+
model: present_model(parsed.model) || present_model(snapshot[:model]) || ParsedUsage::UNKNOWN_MODEL
|
|
119
119
|
)
|
|
120
120
|
end
|
|
121
121
|
|
|
@@ -136,7 +136,7 @@ module LlmCostTracker
|
|
|
136
136
|
|
|
137
137
|
ParsedUsage.build(
|
|
138
138
|
provider: @provider,
|
|
139
|
-
model: snapshot[:model],
|
|
139
|
+
model: snapshot[:model] || ParsedUsage::UNKNOWN_MODEL,
|
|
140
140
|
input_tokens: input,
|
|
141
141
|
output_tokens: output,
|
|
142
142
|
stream: true,
|
|
@@ -148,7 +148,7 @@ module LlmCostTracker
|
|
|
148
148
|
def build_unknown_usage(snapshot)
|
|
149
149
|
ParsedUsage.build(
|
|
150
150
|
provider: @provider,
|
|
151
|
-
model: snapshot[:model],
|
|
151
|
+
model: snapshot[:model] || ParsedUsage::UNKNOWN_MODEL,
|
|
152
152
|
input_tokens: 0,
|
|
153
153
|
output_tokens: 0,
|
|
154
154
|
total_tokens: 0,
|