Model Usage

Name: model-usage
Rating: 78
Author: phuetz

Overview

Track per-model usage, costs, and token consumption to understand spending and optimize model selection.

Quick Queries

Total cost across all sessions

bash

total=0; for f in .codebuddy/sessions/*.json; do
  cost=$(jq -r '.usage.totalCost // 0' "$f" 2>/dev/null)
  total=$(echo "$total + $cost" | bc 2>/dev/null || echo "$total")
done; echo "Total: \$$total"

Cost per model

bash

for f in .codebuddy/sessions/*.json; do
  jq -r '"\(.model // "unknown") \(.usage.totalCost // 0)"' "$f" 2>/dev/null
done | awk '{a[$1]+=$2} END {for(m in a) printf "%-30s $%.4f\n", m, a[m]}' | sort -t'$' -k2 -rn

Token usage per session

bash

for f in .codebuddy/sessions/*.json; do
  jq -r '{file: input_filename, tokens: .usage.totalTokens, cost: .usage.totalCost} | "\(.file)\t\(.tokens)\t$\(.cost)"' "$f" 2>/dev/null
done | column -t -s$'\t'

Daily cost summary

bash

for f in .codebuddy/sessions/*.json; do
  date=$(jq -r '.createdAt // ""' "$f" 2>/dev/null | cut -dT -f1)
  cost=$(jq -r '.usage.totalCost // 0' "$f" 2>/dev/null)
  [ -n "$date" ] && echo "$date $cost"
done | awk '{a[$1]+=$2} END {for(d in a) printf "%s $%.4f\n", d, a[d]}' | sort -r

Optimization Tips

•Use cheaper models (Gemini Flash, Haiku) for simple tasks
•Reserve expensive models (Opus, GPT-4) for complex reasoning
•Monitor token counts — large context windows cost more
•Use MAX_COST env var to set session spending limits
•Check buddy doctor for current model and provider status

Environment Variables

Variable	Description
`MAX_COST`	Session cost limit (default: $10, YOLO: $100)
`GROK_MODEL`	Override default model

Model Cost Reference

Model	Input (per 1M tokens)	Output (per 1M tokens)
grok-3	~$3.00	~$15.00
claude-sonnet-4	~$3.00	~$15.00
gpt-4o	~$2.50	~$10.00
gemini-2.0-flash	~$0.10	~$0.40
ollama (local)	Free	Free

Prices are approximate — check provider dashboards for current rates.