@goodandready/dsh-cost-meter 0.8.28 → 0.8.30
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.
- package/lib/analytics.js +56 -1
- package/lib/client.js +2544 -2379
- package/lib/index.js +1 -0
- package/lib/routes.js +13 -1
- package/package.json +1 -1
package/lib/analytics.js
CHANGED
|
@@ -90,4 +90,59 @@ export function calculateBurnRateAndProjection(turns, totalCostUsd, options = {}
|
|
|
90
90
|
projectedCostUsd,
|
|
91
91
|
confidence,
|
|
92
92
|
}
|
|
93
|
-
}
|
|
93
|
+
}
|
|
94
|
+
|
|
95
|
+
/**
|
|
96
|
+
* Returns estimated context window token limit for model identifier.
|
|
97
|
+
*/
|
|
98
|
+
export function getFallbackContextLength(model) {
|
|
99
|
+
const m = String(model || '').toLowerCase()
|
|
100
|
+
if (m.includes('claude-3-5') || m.includes('claude-3-7') || m.includes('claude-3-opus')) return 200000
|
|
101
|
+
if (m.includes('claude')) return 200000
|
|
102
|
+
if (m.includes('gemini-1.5') || m.includes('gemini-2.0') || m.includes('gemini-exp')) return 1000000
|
|
103
|
+
if (m.includes('gpt-4.1') || m.includes('gpt-4o') || m.includes('o1') || m.includes('o3')) return 128000
|
|
104
|
+
if (m.includes('qwen-2.5') || m.includes('qwq')) return 128000
|
|
105
|
+
if (m.includes('deepseek')) return 128000
|
|
106
|
+
return 128000
|
|
107
|
+
}
|
|
108
|
+
|
|
109
|
+
/**
|
|
110
|
+
* Calculates context window saturation metrics and compaction risk.
|
|
111
|
+
*
|
|
112
|
+
* @param {Array<{ turn: number, uncachedInputTokens?: number, cacheReadTokens?: number }>} turns
|
|
113
|
+
* @param {object} [tariff]
|
|
114
|
+
* @param {object} [route]
|
|
115
|
+
* @returns {{ currentTokens: number, contextLength: number, saturationPercent: number, riskLevel: 'low' | 'moderate' | 'high', nextTurnEstUsd: number }}
|
|
116
|
+
*/
|
|
117
|
+
export function calculateContextMetrics(turns, tariff, route) {
|
|
118
|
+
const lastTurn = (Array.isArray(turns) && turns.length > 0) ? turns[turns.length - 1] : null
|
|
119
|
+
const currentTokens = lastTurn
|
|
120
|
+
? (Number(lastTurn.uncachedInputTokens || 0) + Number(lastTurn.cacheReadTokens || 0))
|
|
121
|
+
: 0
|
|
122
|
+
|
|
123
|
+
const contextLength = (tariff && Number(tariff.contextLength)) || getFallbackContextLength(route?.model)
|
|
124
|
+
const saturationPercent = contextLength > 0
|
|
125
|
+
? Math.min(100, Math.round((currentTokens / contextLength) * 100))
|
|
126
|
+
: 0
|
|
127
|
+
|
|
128
|
+
let riskLevel = 'low'
|
|
129
|
+
if (saturationPercent >= 80) {
|
|
130
|
+
riskLevel = 'high'
|
|
131
|
+
} else if (saturationPercent >= 50) {
|
|
132
|
+
riskLevel = 'moderate'
|
|
133
|
+
}
|
|
134
|
+
|
|
135
|
+
const rates = tariff?.schedule?.base || {}
|
|
136
|
+
const inputRate = Number(rates.input) || 0.14
|
|
137
|
+
const nextTurnEstUsd = currentTokens > 0
|
|
138
|
+
? Number(((currentTokens * inputRate) / 1e6).toFixed(6))
|
|
139
|
+
: 0
|
|
140
|
+
|
|
141
|
+
return {
|
|
142
|
+
currentTokens,
|
|
143
|
+
contextLength,
|
|
144
|
+
saturationPercent,
|
|
145
|
+
riskLevel,
|
|
146
|
+
nextTurnEstUsd,
|
|
147
|
+
}
|
|
148
|
+
}
|