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.
Typical workloads
Custom workload
Clicking a preset fills the calculator below with that workload. Every field stays editable — change a number or slider and the workload counts as custom again.
Workload: 100.00M in · 20.00M out per month
Models to compare (3 / 3)
Projected monthly spend
Switching from OpenAI: GPT-6.1 Terra Pro to OpenAI: GPT-6.1 Terra Pro (batch) saves $1,029.00 per year.
OpenAI: GPT-6.1 Terra Pro (batch) OpenAI$85.75/mo−$85.75/mo
OpenAI: GPT-6.1 Terra Pro baselineOpenAI$171.50/mo—
OpenAI: GPT-6.1 Terra OpenAI$171.50/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 29, 2026, 02:42 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. Workload presets only fill these fields with typical volumes — they add no data of their own.
Help