AI API Cost Calculator
Estimate cost per request, day, and month across Claude, GPT-5, Gemini, and friends. Toggle prompt caching and batch mode to see real-world savings. Pricing snapshot: 2026-04-29.
Workload
Model A
AnthropicEffective input$3.00/1M
Effective output$15.00/1M
Per request$0.0135
Per day$13.50
Per month$405.00
Model B
OpenAIEffective input$5.00/1M
Effective output$20.00/1M
Per request$0.0200
Per day$20.00
Per month$600.00
Model A is $195.00/mo cheaper
Weekly cost-optimization playbooks.
Get the techniques we use to cut AI bills 50 to 80% - prompt caching patterns, batch pipelines, model routing, and eval-driven downsizing. Free, no fluff.
Get Smarter About AI Dev
New tutorials, open-source projects, and deep dives on coding agents - delivered weekly.
One email per weekReal code, not theoryFree forever
Cut your bill further
Blog
Prompt caching in production with the Claude API
90% off cached reads, when to invalidate, and the gotchas that bite at scale.
Blog
Tool use patterns that keep token spend low
How to structure tool calls so you do not pay for re-sent context every turn.
Blog
Cost tape: line-item observability for AI calls
Wire spend per route, per user, per skill so you can find the leak.
YouTube
Cost-optimization videos on Developers Digest
Walkthroughs: caching, batch APIs, model cascades, and routing.
FAQ
How is AI API cost calculated?+
(input tokens / 1M) * input price + (output tokens / 1M) * output price, multiplied by requests per day. Output tokens are typically 3 to 5x more expensive than input.
How much does prompt caching save?+
Anthropic gives a 90% discount on cached input tokens. Cache writes cost 25% more than base input, so caching pays off when context is reused enough times. On stable system prompts you can see 50 to 80% total savings.
What is batch mode?+
An async API that returns results within 24 hours at 50% off both input and output. Ideal for evals, backfills, summarization, and offline pipelines.
Are prices current?+
Snapshot from 2026-04-29. Verify against provider pricing pages before committing to a contract.
Pricing as of 2026-04-29. Always verify against provider pricing pages.