⚫ Grok Token Calculator
What a prompt costs on xAI's Grok. It is the simplest pricing on this site: one rate, no tiers, no promotional expiry, and a narrower gap between input and output than most rivals.
The simplest pricing here, and what that is worth
Grok's pricing is notable mainly for what it does not have. There are no tiers, no promotional expiry dates, no prompt-length thresholds and no separate cached-input rate published alongside the model. One input price, one output price, and a 500,000-token context window.
Current prices
| Model | Input | Output | Cache read |
|---|---|---|---|
| Grok 4.7 | $2 | $6 | — |
A narrower output multiple
Grok 4.7 charges $2 input and $6 output, a multiple of three. Most rivals charge four to eight times input for output, so on a generation-heavy workload — long drafts, code, extended reasoning — Grok's relative position improves the more the model writes. On a retrieval-heavy workload with short answers, the advantage disappears.
The context window is the headline feature
500,000 tokens is large enough to hold a substantial codebase or a book. The thing to remember is that a large context window is a capability, not a budget: filling it costs $1 per call at $2 per million, and if you fill it on every turn of a conversation the arithmetic in the conversation cost calculator applies with force.
Image and video are priced separately
Grok Imagine charges from $0.02 per image and $0.02 per second of video, which is pay-as-you-go with no subscription. For occasional image generation that undercuts a Midjourney plan considerably; at high volume the subscription wins. The image cost calculator shows where the lines cross.
What simple pricing is actually worth
Predictability. There is no tier to get wrong, no threshold to cross accidentally, and no date after which your forecast breaks. For a small team that cannot afford to monitor five providers' pricing pages, that is a real advantage even when the headline rate is not the lowest available.
Prices come from xAI models and pricing and were checked on 2026-10-26. AI pricing changes frequently — confirm against the provider before committing to a budget.
If this is one step in a larger decision, the ai token calculator is usually the next question, and the deepseek token cost is worth a look while you are here.
Frequently Asked Questions (FAQs)
How much does the Grok API cost?
Grok 4.7 was $2 per million input tokens and $6 output on 2026-10-26, with a 500,000-token context window and no tier structure.
Does Grok offer cached input pricing?
No separate cache read rate is published alongside the model on xAI's model page, so this calculator does not model a caching discount for it. Check the provider's full pricing page if caching matters to your design.
How much does Grok Imagine cost?
Images start at $0.02 each at 1K or 2K, and video at $0.02 per second — so a ten-second clip is about $0.20. It is pay-per-use with no subscription.
Is Grok cheaper than GPT or Claude?
At the same input price as Sonnet 5.5 and GPT-6.1 Sol, Grok's advantage is on output, where it charges three times input rather than five. That makes it relatively better on generation-heavy work and neutral elsewhere.
What can I do with a 500,000-token context?
Hold a large codebase, a long book or a substantial document set in a single call. Remember that filling it costs about $1 per call, and resending it each turn of a conversation multiplies that quickly.
Does Grok have promotional pricing?
Not at the time of checking. Unlike several Gemini models, there was no published expiry date or scheduled increase, which makes forecasting more straightforward.
Are these prices current?
They were read from xAI's model documentation on 2026-10-26. The page links to a fuller pricing page, so check there for tiers or discounts not covered on the model listing.