mirror of
https://github.com/getcompanion-ai/co-mono.git
synced 2026-04-21 01:01:42 +00:00
feat(ai): Add cost tracking to LLM implementations
- Track input/output token costs for all providers - Calculate costs based on Model pricing information - Include cost information in AssistantMessage responses - Add Usage interface with detailed cost breakdown - Implement calculateCost utility function for cost calculations
This commit is contained in:
parent
f9d688d577
commit
550da5e47c
6 changed files with 61 additions and 14 deletions
|
|
@ -5,6 +5,7 @@ import {
|
|||
type GenerateContentParameters,
|
||||
GoogleGenAI,
|
||||
} from "@google/genai";
|
||||
import { calculateCost } from "../models.js";
|
||||
import type {
|
||||
AssistantMessage,
|
||||
Context,
|
||||
|
|
@ -13,9 +14,9 @@ import type {
|
|||
Message,
|
||||
Model,
|
||||
StopReason,
|
||||
TokenUsage,
|
||||
Tool,
|
||||
ToolCall,
|
||||
Usage,
|
||||
} from "../types.js";
|
||||
|
||||
export interface GoogleLLMOptions extends LLMOptions {
|
||||
|
|
@ -97,11 +98,12 @@ export class GoogleLLM implements LLM<GoogleLLMOptions> {
|
|||
let thinking = "";
|
||||
let thoughtSignature: string | undefined;
|
||||
const toolCalls: ToolCall[] = [];
|
||||
let usage: TokenUsage = {
|
||||
let usage: Usage = {
|
||||
input: 0,
|
||||
output: 0,
|
||||
cacheRead: 0,
|
||||
cacheWrite: 0,
|
||||
cost: { input: 0, output: 0, cacheRead: 0, cacheWrite: 0, total: 0 },
|
||||
};
|
||||
let stopReason: StopReason = "stop";
|
||||
let inTextBlock = false;
|
||||
|
|
@ -179,6 +181,13 @@ export class GoogleLLM implements LLM<GoogleLLMOptions> {
|
|||
(chunk.usageMetadata.candidatesTokenCount || 0) + (chunk.usageMetadata.thoughtsTokenCount || 0),
|
||||
cacheRead: chunk.usageMetadata.cachedContentTokenCount || 0,
|
||||
cacheWrite: 0,
|
||||
cost: {
|
||||
input: 0,
|
||||
output: 0,
|
||||
cacheRead: 0,
|
||||
cacheWrite: 0,
|
||||
total: 0,
|
||||
},
|
||||
};
|
||||
}
|
||||
}
|
||||
|
|
@ -203,6 +212,9 @@ export class GoogleLLM implements LLM<GoogleLLMOptions> {
|
|||
thoughtSignature = Buffer.from(signature).toString("base64");
|
||||
}
|
||||
|
||||
// Calculate cost
|
||||
calculateCost(this.model, usage);
|
||||
|
||||
// Usage metadata is in the last chunk
|
||||
// Already captured during streaming
|
||||
|
||||
|
|
@ -227,6 +239,7 @@ export class GoogleLLM implements LLM<GoogleLLMOptions> {
|
|||
output: 0,
|
||||
cacheRead: 0,
|
||||
cacheWrite: 0,
|
||||
cost: { input: 0, output: 0, cacheRead: 0, cacheWrite: 0, total: 0 },
|
||||
},
|
||||
stopReason: "error",
|
||||
error: error instanceof Error ? error.message : String(error),
|
||||
|
|
|
|||
Loading…
Add table
Add a link
Reference in a new issue