Pricing
Updated:
Rates are shown per 1M tokens. Long context pricing applies above the input threshold shown for each model.
Regular discount — 50%Now — 82% until
| Model | Short context | Long context | ||||
|---|---|---|---|---|---|---|
| Input | Cached | Output | Input | Cached | Output | |
| grok-4.5 Long context ≥ 200k tokens | $0.36 | $0.054 | $1.08 | $0.72 | $0.108 | $2.16 |
| grok-4.6 Long context ≥ 200k tokens | $0.36 | $0.09 | $1.08 | $0.72 | $0.18 | $2.16 |
| grok-4.7 Long context ≥ 200k tokens | $0.36 | $0.09 | $1.08 | $0.72 | $0.18 | $2.16 |
How the price is worked out
Billing is measured in tokens. Every call carries two counters: the input tokens you send and the output tokens the model generates. Both are billed at the published rates for the selected model.
Input tokens
Everything you send: the system prompt, the conversation history, any retrieved context, and the user's own message.
Output tokens
Everything the model generates, reasoning tokens included. Priced 3× higher than input.
Cached input
A repeated prompt prefix lands in the cache and is billed about 4× cheaper than regular input.
What a token is
A token is the basic unit of text a model works with. For English one token is roughly 4 characters, or about three quarters of a word. Treat that as a guide rather than a rule: the real count depends on the model, the language, and the content. Cyrillic, Chinese characters, code, and rare words split into more tokens for the same amount of text.
- “Hello, world!”
- about 4 tokens
- A 1,000-word English article
- about 1,300 tokens
- A typical source file
- about 500–2,000 tokens
You never have to count them by hand: every API response returns the exact input and output totals, and the charge is taken from those.
A worked example
Given:
- Active users
- 300
- Input tokens per user per day
- 12,000
- Output tokens per user per day
- 3,000
- Days in the month
- 30
- Model
- grok-4.7
- Input price per 1M
- $2.00 → $0.36
- Output price per 1M
- $6.00 → $1.08
Find:
Cost per month
Solution:
Input tokens per month
300 × 12,000 × 30108,000,000
Output tokens per month
300 × 3,000 × 3027,000,000
At official rates
108M × $2.00 + 27M × $6.00$378.00
Through GrokAPI
108M × $0.36 + 27M × $1.08$68.04
Answer:
$309.96 per month · $3,719.52 per year
The arithmetic is the same at any volume: divide the token count by 1,000,000, multiply by the per-1M rate, then add input and output together. Work out your own load
Frequently asked questions
How much does the Grok API cost?
Grok 4.7 is $0.36 per 1M input tokens and $1.08 per 1M output tokens — 82% below the official xAI rate of $2.00 and $6.00. Repeated prompt prefixes are billed at $0.09 per 1M.
Is there a subscription or a monthly minimum?
No. No subscription, no recurring fee, and no monthly minimum. You top up a single balance and it is drawn down only by the tokens the API actually processes, so an idle month costs nothing.
How do I top up the balance?
With cryptocurrency. Create an account and open the balance section — it lists the supported currencies and networks, along with the address to transfer to. Check the network before sending: a transfer on an unsupported one may not be recoverable.
When does long context pricing apply?
At or above 200k input tokens in one request: $0.72 per 1M input and $2.16 per 1M output. Input includes regular input, cache reads and cache writes; output does not count towards the threshold.
Can I try it before paying?
Yes. A new account gets $0.25 in test credit right after sign-up — enough to verify the connection and send your first requests before topping up.
Ready to connect Grok models?
Open your account, add balance, and create an API key.