Speakr is a personal, self-hosted web application designed for transcribing audio recordings
Python
"""
Token usage tracking model for monitoring LLM API consumption.
"""
from datetime import datetime, date
from src.database import db
class TokenUsage(db.Model):
"""Daily token usage aggregates per user per operation type."""
__tablename__ = 'token_usage'
id = db.Column(db.Integer, primary_key=True)
user_id = db.Column(db.Integer, db.ForeignKey('user.id'), nullable=False)
date = db.Column(db.Date, nullable=False, default=date.today)
operation_type = db.Column(db.String(50), nullable=False)
# Token counts (from API response.usage)
prompt_tokens = db.Column(db.Integer, default=0)
completion_tokens = db.Column(db.Integer, default=0)
total_tokens = db.Column(db.Integer, default=0)
# Prompt-cache counts (subset of prompt_tokens). cached_tokens are prompt
# tokens served from a prefix cache (cache reads); cache_write_tokens are
# tokens written into the cache on this call. Providers expose these under
# usage.prompt_tokens_details.cached_tokens / usage.cache_creation_input_tokens.
cached_tokens = db.Column(db.Integer, default=0)
cache_write_tokens = db.Column(db.Integer, default=0)
# Cost tracking (OpenRouter provides this)
cost = db.Column(db.Float, default=0.0)
# Request count for this day/operation
request_count = db.Column(db.Integer, default=0)
# Model info
model_name = db.Column(db.String(100), nullable=True)
created_at = db.Column(db.DateTime, default=datetime.utcnow)
updated_at = db.Column(db.DateTime, default=datetime.utcnow, onupdate=datetime.utcnow)
# Relationships
user = db.relationship('User', backref=db.backref('token_usage', lazy='dynamic', cascade='all, delete-orphan'))
__table_args__ = (
db.UniqueConstraint('user_id', 'date', 'operation_type', name='uq_user_date_op'),
db.Index('idx_token_user_date', 'user_id', 'date'),
)
def __repr__(self):
return f'<TokenUsage {self.user_id} {self.date} {self.operation_type}: {self.total_tokens}>'