300+ Tools CoveredSource Data Updated Weeklydates

Compare OpenAI, Anthropic and Gemini API costs

Full comparison free for a limited time

12-month cost model · PDF report + Excel + CSV

Your workload

Models to compare (up to 4)

≈ 80% per year, compoundedApplied to input tokens and output tokens; the selected models stay the same.

or see a sample comparison (PDF with a fictional workload, opens in a new tab)

Uses verified public pricing. Cost model, not a performance benchmark.

Advanced assumptions
What's included / excluded

Estimated cost of the selected models for one monthly token workload, with prompt caching, Batch and long prompts. Performance, throughput and latency are not assumed to be equal.

Included

  • Input and output tokens at each model's rates per 1M tokens
  • Cached input, cache writes and Gemini's cache storage
  • Batch processing
  • Prompts above a model's standard price band

Not included

  • Flex, Priority and other processing modes
  • Tool use such as web search, grounding or code execution
  • Taxes, credits and negotiated discounts
  • Model quality, speed or capability equivalence

Prices modeled: GPT-6 Astra · OpenAI API; GPT-6.1 Sol · OpenAI API; GPT-6.1 Terra · OpenAI API; GPT-6 Luna · OpenAI API; Claude Fable 5.1 · Claude API; Claude Opus 5.5 · Claude API; Claude Sonnet 5.5 · Claude API; Claude Haiku 5.5 · Claude API; Gemini 4 Argon · Gemini API; Gemini 3.1 Pro Preview · Gemini API; Gemini 3.5 Flash-Lite · Gemini API.

Full comparison free for a limited time.

Estimates are based on publicly available vendor pricing and the assumptions shown. Actual invoices may differ because of negotiated discounts, taxes, regional pricing, commitments, overages or product configuration.