TokenTally

xAI frontier models

Grok 4.20 Beta token costs

Model the $2 in / $6 out pricing for Grok 4.20, including cache-hit discounts and the 2M-token context window.

Last pricing check: Mar 13, 2026

$2.00 per 1M prompt tokens$6.00 per 1M completion tokens2,000,000 token context

Why teams choose this model

Multi-agent research and planning
High-stakes ops copilots
Enterprise orchestration bots

Scenario planning

Realistic cost examples

Numbers use Grok 4.20 Beta (Reasoning) pricing

Agentic research pod

Analysts spin up tool-heavy Grok 4.20 agents for deep dives with cacheable system prompts.

Per request

$0.0128

Per month

$30.72

Tokens sent

8,640,000

2200 prompt tokens1400 completion tokens2,400 requests/mo

Incident command center

24/7 ops copilots summarize telemetry, generate action plans, and post structured updates.

Per request

$0.0108

Per month

$38.88

Tokens sent

10,800,000

1800 prompt tokens1200 completion tokens3,600 requests/mo

Product strategy brief

Leadership asks Grok for quarterly briefs that stitch together docs, dashboards, and X search pulls.

Per request

$0.0154

Per month

$13.86

Tokens sent

3,870,000

2600 prompt tokens1700 completion tokens900 requests/mo

Compare with

FAQs

How do cache-hit prices factor in?

When Grok 4.20 recognizes a cached prompt prefix, the input rate drops to $0.20/M tokens. TokenTally stores that discount so you can compare best vs. default cases.

Does the 2M context increase costs?

Only if you fill it. The scenarios keep prompts under 3K tokens, but you can paste any size prompt into the calculator to see real-time totals.

Are tool invocations billed separately?

Yes—xAI charges $5 per 1K calls for Web/X search or code tools. Those fees stack on top of the token totals shown here.

Pricing sources