Short answer: Grok 4.7, xAI's flagship model, costs $2 per million input tokens, $0.50 per million cached input tokens and $6 per million output tokens, with a 500k-token context window. A prompt of 200k tokens or more is billed at double those rates for the whole request. Grok 4.3 and the Grok 4.20 models cost less ($1.25 in, $2.50 out) and accept up to 1M tokens. Prices are xAI's list prices, checked November 6, 2026.
Grok API prices per million tokens
These are the standard rates for prompts under 200k tokens, from xAI's model docs. They are list prices in US dollars and do not include tax. If xAI's page shows a different number, trust that page.
| Model | Context window | Input | Cached input | Output |
|---|---|---|---|---|
| grok-4.7 (flagship) | 500k | $2.00 | $0.50 | $6.00 |
| grok-4.6 | 500k | $2.00 | $0.50 | $6.00 |
| grok-4.5 | 500k | $2.00 | $0.30 | $6.00 |
| grok-4.3 | 1M | $1.25 | $0.20 | $2.50 |
| grok-4.20-0309-reasoning | 1M | $1.25 | $0.20 | $2.50 |
| grok-4.20-0309-non-reasoning | 1M | $1.25 | $0.20 | $2.50 |
| grok-4.20-multi-agent-0309 | 1M | $1.25 | $0.20 | $2.50 |
| grok-build-0.1 | 256k | $1.00 | $0.20 | $2.00 |
How long-context pricing works
Every text model has a second price tier. Once a prompt reaches 200k tokens, the higher rate applies to all tokens in that request, not only the tokens past 200k.
For Grok 4.7 and Grok 4.6 the long-context rates are $4 input, $1 cached input and $12 output per million tokens. Grok 4.5 is $4, $0.60 and $12. Grok 4.3 and the Grok 4.20 models are $2.50, $0.40 and $5. grok-build-0.1 is $2, $0.40 and $4.
A worked example: a 250k-token prompt to Grok 4.7 costs $1.00 for input (250k at $4 per million). Split into two 125k-token prompts, the same input costs $0.50. If your job allows it, keeping prompts under 200k halves the bill.
Grok context windows
Grok 4.7, 4.6 and 4.5 accept up to 500k tokens. Grok 4.3 and the Grok 4.20 models accept up to 1M tokens. grok-build-0.1 accepts 256k. According to xAI's release notes, Grok 4.7 and 4.6 take text and image input and return text only.
A larger window is useful for long documents, codebases and agent runs that keep a lot of history. It is also where the long-context price starts, so check whether a job really needs more than 200k tokens before you send them.
Which Grok model to pick
Grok 4.7 is the default for hard work: xAI calls it the most capable model it has built and recommends it for code and chat. Reasoning effort can be set to low, medium, high (the default) or xhigh. Higher settings usually use more output tokens, so they cost more per task.
Grok 4.3 or Grok 4.20 fit high-volume jobs and very long inputs, at less than half the flagship's output price and with a 1M window.
Grok 4.7 Fast costs twice the standard rate (1.5 times for long-context requests). It is only available in Cursor and Grok Build, not as a model on the public xAI API.
Grok 4.6 and 4.7 on Amazon Bedrock and Microsoft Foundry
Amazon Bedrock: Grok 4.6 became generally available on Bedrock on August 18, 2026, a week after it launched. Model IDs are xai.grok-4.6 on bedrock-mantle (in-Region in US West, Oregon) and the cross-Region profiles us.xai.grok-4.6 and global.xai.grok-4.6 on bedrock-runtime. AWS lists $2.00 per million input tokens through global cross-Region inference and $2.20 in-Region. It works with the Responses, Chat Completions and Converse APIs.
Microsoft Foundry: Grok 4.6 arrived in public preview on August 26, 2026. Grok 4.7 has been generally available in Foundry Models since October 6, 2026. Global Standard deployments cost $2 input, $0.50 cached and $6 output per million tokens. US Data Zone deployments cost $2.20, $0.55 and $6.60.
Watermarking from November 7, 2026
From November 7, 2026, text output from grok-4.7, grok-4.6, grok-4.5, grok-4.3 and the grok-4.20 models carries an invisible statistical watermark by default. Each response has an x-watermark header, and true means that output was watermarked. Team admins can turn it off per model in the console, but only after confirming a legal exception applies. Some models may stay locked on. Read our watermark news post for the details.
Image, video and voice prices
| Model | Price |
|---|---|
| grok-imagine-image-2.0 | $0.04 per image |
| grok-imagine-image | $0.02 per image |
| grok-imagine-video-1.5 | $0.08 per second |
| grok-imagine-video-1.5-lite | $0.02 per second |
| grok-voice-3.0 (speech to speech) | $0.08 per minute |
| Speech to text | $0.10 per hour (REST), $0.20 per hour (streaming) |
| Text to speech | $15 per 1M characters |
Grok API vs Grok Bot and the Grok apps
The API is billed per token to developers. Grok Bot and the Grok apps are paid through The X Subscription, where usage comes from a weekly allowance. Plan prices are in how much Grok Bot costs, and the recent change to allowances is in Grok Bot usage limits.
Want to see what people build on top of these models? Browse verified shared Grok Bots or start from a Grok template.
Before adding any of these, read Using a Grok Bot safely.
References
- xAI docs: Models and pricing — prices and context windows, checked November 6, 2026
- xAI docs: Release notes — Grok 4.6, Grok 4.7 and Grok 4.7 Fast
- xAI docs: Watermarking
- AWS: Grok 4.6 in Amazon Bedrock
- xAI: Grok 4.6 on Amazon Bedrock
- Microsoft: Grok 4.7 on Microsoft Foundry