Grok 4.20 (reasoning) API cost calculator

Grok 4.20 (reasoning) (Grok (xAI)) is priced at $1.25 per million input tokens, $0.2 for cached input, and $2.50 per million output tokens. Verified 2026-09-17 against Grok (xAI)'s pricing page.

Requests with 200k or more input tokens are billed at twice these rates.

What Grok 4.20 (reasoning) costs for common workloads

WorkloadPer requestPer dayPer month
Support chatbotShort turns with a reused system prompt — 3k in, 400 out, 60% cached.$0.0029$14.30$429
RAG over documentsRetrieved context dominates — 12k in, 500 out, 30% cached.$0.012$18.70$561
Coding agentGrowing context every turn — 20k in, 1.5k out, 80% cached.$0.012$35.85$1,076
Batch classificationTiny prompts at volume through the Batch API — 800 in, 20 out.$0.0010$210$6,300

Its blast radius when an agent loops

The coding-agent workload above at 30 requests a minute — $21.51 an hour — priced at each point a team usually finds out:

Hourly alert

$21.51

Next morning

$516

After a weekend

$1,291

Monthly invoice

$15,487

Get the alert, not the invoice — start free

Run your own numbers on Grok 4.20 (reasoning)

Other Grok (xAI) models