Cost Tracking

The CostTrackingService monitors token usage and computes estimated costs for every API call. It maintains pricing tables for 30+ models across 9 providers and produces per-conversation, per-provider, and aggregate summaries.

How it works

Every API call to an LLM provider consumes tokens. CodeBuddy tracks:

• Input tokens — The prompt and context sent to the model
• Output tokens — The model’s response
• Estimated cost — Computed with 6-decimal-place precision using per-model pricing tables

Cost is calculated as:

$$\text{cost} = \frac{\text{input tokens} \times \text{input price}}{1{,}000{,}000} + \frac{\text{output tokens} \times \text{output price}}{1{,}000{,}000}$$

Viewing costs

Cost information is displayed at three levels:

• Per message — Token count shown below each response
• Per conversation — Cumulative cost for the current thread
• Aggregate summary — Total cost broken down by provider and model

Setting cost limits

Prevent unexpected charges with a per-task cost limit:

{
  "codebuddy.costLimit": 1.0
}

When the limit is reached, CodeBuddy pauses and asks if you want to continue.

Pricing by provider

Approximate per-million-token pricing (mid-2025):

Provider	Model	Input	Output
Anthropic	Claude Sonnet 4	$3.00	$15.00
Anthropic	Claude Opus 4	$15.00	$75.00
Anthropic	Claude Haiku	$0.25	$1.25
OpenAI	GPT-4o	$2.50	$10.00
OpenAI	GPT-4o-mini	$0.15	$0.60
OpenAI	o3-mini	$1.10	$4.40
Google	Gemini 2.5 Pro	$1.25	$10.00
Google	Gemini 2.5 Flash	$0.15	$0.60
Groq	Llama 3.3 70B	$0.59	$0.79
DeepSeek	DeepSeek Chat	$0.27	$1.10
DeepSeek	DeepSeek Reasoner	$0.55	$2.19
Qwen	Qwen Plus	$0.80	$2.00
xAI	Grok	$5.00	$15.00
Ollama	Local models	Free	Free

When a model isn’t in the pricing table, a conservative fallback of $3.00 / $15.00 per million tokens is used.

Cost summary API

The service exposes structured cost data:

interface ICostSummary {
  totals: {
    inputTokens: number;
    outputTokens: number;
    estimatedCost: number;
    requestCount: number;
  };
  providers: Array<{
    provider: string;
    model: string;
    inputTokens: number;
    outputTokens: number;
    estimatedCost: number;
    requestCount: number;
  }>;
  conversations: Array<{
    threadId: string;
    provider: string;
    model: string;
    inputTokens: number;
    outputTokens: number;
    estimatedCost: number;
    requestCount: number;
  }>;
}

Tips for reducing costs

• Use smaller models for simple tasks — GPT-4o-mini or Gemini Flash for file renames, formatting, simple edits
• Use local models — Ollama for tasks that don’t need frontier intelligence
• Set project rules — Clear rules reduce unnecessary iterations and context growth
• Set a cost limit — Use codebuddy.costLimit as a safety net
• Monitor context growth — Long conversations accumulate tokens. Start new conversations for unrelated tasks