Cost audit

Workload Cost Audit

Describe your monthly workload or paste a sample prompt, then compare projected spend across OpenAI, Anthropic, Google Gemini, Meta Llama (as served by Groq and others) and DeepSeek.

Input (M tokens / mo)Output (M tokens / mo)

Prompt cache hit rate — 30%Batch processing share — 0%

Workload: 100.00M in · 20.00M out per month

Models to compare (3 / 3)

Projected monthly spend

OpenAI: GPT-6.1 Sol Pro baselineOpenAI$343.00/mo—
OpenAI: GPT-6.1 Sol OpenAI$343.00/mo—
Anthropic: Claude Sonnet 5.5 Anthropic$346.00/mo+$3.00/mo

Cache hits use each model's published cache-read price where available. Batch share gets 50% off for OpenAI, Anthropic and Google. Free audits compare up to 3 models PRO

As of: Oct 1, 2026, 12:44 PM UTCSource: OpenRouter Models API Method: Projected spend multiplies the entered workload by published prices. Sample prompts use an approximate four characters per input token; cache and batch assumptions are shown above.