xAI Grok API Pricing: Text, Image, Voice & Video (2026)
Last updated:
How much does xAI Grok cost? Below 200K input tokens, Grok Build 0.1 is $1 input / $0.2 cached / $2 output per 1M tokens, Grok 4.3 and the dated Grok 4.20 SKUs are $1.25 input / $0.2 cached / $2.5 output, and Grok 4.7 is $2 input / $0.5 cached / $6 output. At 200K input tokens or more, xAI applies the higher long-context tier shown below.
How much does each xAI Grok model cost per million tokens?
| Try it | |||||||
|---|---|---|---|---|---|---|---|
Grok 4.7 | xAI | Mid | 500,000 | $2.00 | $0.50 | $6.00 | Try API → |
Grok 4.3 | xAI | Mid | 1M | $1.25 | $0.20 | $2.50 | Try API → |
Grok 4.20 0309 Non-Reasoning | xAI | Mid | 1M | $1.25 | $0.20 | $2.50 | Try API → |
Grok 4.20 0309 Reasoning | xAI | Mid | 1M | $1.25 | $0.20 | $2.50 | Try API → |
Grok 4.20 Multi-Agent 0309 | xAI | Mid | 1M | $1.25 | $0.20 | $2.50 | Try API → |
Grok 4.6 | xAI | Mid | 500,000 | $2.00 | $0.50 | $6.00 | Try API → |
Grok Build 0.1 | xAI | Low | 256,000 | $1.00 | $0.20 | $2.00 | Try API → |
- Grok 4.7xAIMid
- Input
- $2.00
- Cached
- $0.50
- Output
- $6.00
- Grok 4.3xAIMid
- Input
- $1.25
- Cached
- $0.20
- Output
- $2.50
- Grok 4.20 0309 Non-ReasoningxAIMid
- Input
- $1.25
- Cached
- $0.20
- Output
- $2.50
- Grok 4.20 0309 ReasoningxAIMid
- Input
- $1.25
- Cached
- $0.20
- Output
- $2.50
- Grok 4.20 Multi-Agent 0309xAIMid
- Input
- $1.25
- Cached
- $0.20
- Output
- $2.50
- Grok 4.6xAIMid
- Input
- $2.00
- Cached
- $0.50
- Output
- $6.00
- Grok Build 0.1xAILow
- Input
- $1.00
- Cached
- $0.20
- Output
- $2.00
Sponsored links may earn us a commission at no extra cost to you. Affiliate status never changes model ordering.
All product names, logos, and brands are property of their respective owners and are used for identification purposes only.
Last synced:
New · Grok 4.7
New flagship, unchanged Grok 4.6 token rates
xAI lists Grok 4.7 at $2 input / $0.5 cached / $6 output per 1M tokens below 200K prompt tokens and $4 input / $1 cached / $12 output at or above that threshold—the same rates as Grok 4.6. It accepts text and images, supports 500K context, offers four reasoning levels, and is live through the public xAI API. Read the Grok 4.7 pricing and benchmark analysis →
Labs status: Grok 4.6 already scored 49/49 in the maintained deterministic suite. Grok 4.7 has no published AI Pricing Guru result because the maintained OpenRouter transport did not list the exact route at our September 21 check. xAI's launch benchmarks stay separate until the exact model can complete the fixed 49-task run with full responses and billed spend evidence.
Retires November 2, 2026
grok-imagine-image-quality redirects to Image 2.0 low
xAI's 60-day notice began September 2. After retirement, requests to grok-imagine-image-quality keep the same request and response shape but are served by grok-imagine-image-2.0 with quality: "low". The 1K output rate falls from $0.05 to $0.04 per image; 2K falls from $0.07 to $0.06. The older grok-imagine-image 1.0 slug is not affected.
Read the migration and cost analysis →
Labs status: this image-only endpoint cannot be inserted honestly into the text-answer Cost-per-Task leaderboard. A valid retirement test needs pinned pre/post slugs, fixed generation and editing fixtures, repeated outputs, blind acceptance labels, latency, safety failures, and billed cost per accepted image.
New · Grok Bot beta
Plan access, not a new API model or standalone rate
xAI launched Grok Bot for SuperGrok Heavy, Cursor Ultra, and Cursor Teams Premium accounts. It includes weekly usage and can continue on eligible accounts through on-demand model-and-token billing, but xAI publishes no per-Bot, cloud-computer, connector, or routine fee. The product manages model selection and failover, so we did not create a fake Grok Bot row in the API table. Read the Grok Bot pricing and rollout analysis →
New: Grok Voice Think Fast 2.0
xAI now lists grok-voice-think-fast-2.0 at $0.08 per audio minute ($4.80/hour), versus $0.05/minute ($3.00/hour) for 1.0. The grok-voice-latest alias moves to 2.0 on August 5, 2026.
Read the migration and pricing analysis →
Think Fast 2.0 is audio-priced and cannot be scored honestly in the current text-only Labs leaderboard; the launch analysis records the required speech-to-speech harness.
New · September 17, 2026
Grok Voice Transcribe 2.0 is live at the 1.0 rates
xAI describes grok-voice-transcribe-2.0 as its best transcription model, but keeps grok-voice-transcribe-1.0 as the default when the model parameter is omitted. Both versions use the same published REST and WebSocket streaming rates.
Read the Transcribe 2.0 migration and cost analysis →
| Model | REST / hour | Streaming / hour | Default |
|---|---|---|---|
| Grok Voice Transcribe 1.0 | $0.10 | $0.20 | Yes |
| Grok Voice Transcribe 2.0 | $0.10 | $0.20 | No |
Rates are USD per audio hour and come from xAI's official Voice API rate card.
New: Grok Imagine Image 2.0 auto quality
xAI now defaults grok-imagine-image-2.0 to auto quality. Auto currently bills generation at low quality and editing at medium quality. Editing accepts up to 5 source images, and generation/editing now support the new 21:9 and 5:2 aspect ratios.
Read the billing and migration analysis →
| Model | 1K output | 2K output | Input image |
|---|---|---|---|
| Grok Imagine Image | $0.02 | $0.02 | $0.002 |
| Grok Imagine Image 2.0 | $0.04 low / $0.06 medium | $0.06 low / $0.08 medium | $0.010 |
| Grok Imagine Image Quality | $0.05 | $0.07 | $0.010 |
Rates are USD per output or input image and come from xAI's official Imagine rate card.
Labs status: the current 49-task leaderboard evaluates text answers, not visual outputs. An honest image study needs a pinned model and quality, fixed text/editing prompts, licensed reference images, repeated generations, human acceptance labels, latency, retries, safety failures, exact billed cost, and cost per accepted asset.
New: Grok Imagine Video 1.5 modes and 1080p
Grok Imagine Video 1.5 now supports text-to-video, image-to-video, and reference-to-video. Text- and image-to-video can render native 1080p; reference-to-video is capped at 720p. Read the capability and cost analysis →
| Model | 480p / sec | 720p / sec | 1080p / sec |
|---|---|---|---|
| Grok Imagine Video | $0.05 | $0.07 | — |
| Grok Imagine Video 1.5 | $0.08 | $0.14 | $0.25 |
Rates are USD per generated second and come from xAI's official public model metadata.
Key facts about xAI Grok pricing
- Grok 4.7: $2 input / $0.5 cached / $6 output per 1M below 200K prompt tokens; $4 input / $1 cached / $12 output at 200K or more.
- Grok 4.6: $2 input / $0.5 cached / $6 output per 1M below 200K input tokens; $4 input / $1 cached / $12 output at 200K or more.
- Grok 4.5: $2 input / $0.3 cached / $6 output per 1M below 200K input tokens; $4 input / $0.6 cached / $12 output at 200K or more.
- Grok 4.3 and dated Grok 4.20 SKUs: $1.25 input / $0.2 cached / $2.5 output below 200K; $2.5 input / $0.4 cached / $5 output at 200K or more.
- Grok Build 0.1: $1 input / $0.2 cached / $2 output below 200K; $2 input / $0.4 cached / $4 output at 200K or more.
- Context: 500K for Grok 4.7, Grok 4.6, and Grok 4.5; 1M for Grok 4.3 and dated Grok 4.20; and 256K for Grok Build 0.1.
- Grok Bot: no public API model ID or standalone rate; billing follows the managed serving model.
- Grok Voice Transcribe 2.0: $0.10 per REST audio hour and $0.20 per streaming audio hour; 1.0 remains the default.
- Imagine retirement:
grok-imagine-image-qualityredirects to Image 2.0 low on November 2, 2026, lowering the output rate by $0.01 per image.
Why choose xAI Grok over OpenAI or Anthropic?
xAI's current API catalog spans Grok 4.7 as its flagship, a coding-focused Grok Build route, Grok 4.3, Grok 4.5, Grok 4.6, and dated Grok 4.20 reasoning, non-reasoning, and multi-agent endpoints. The direct pricing table is the right comparison for software you build; Grok Bot is a plan-gated application with managed routing and should be evaluated by completed-work cost instead.
Grok 4.3 and the dated Grok 4.20 SKUs list 1M context, Grok 4.7, Grok 4.6, and Grok 4.5 list 500K, and Grok Build 0.1 lists 256K. Every current row also has a higher tier beginning at 200K input tokens, so long-context budgets should use the structured tier rather than multiplying the short-context price.
When xAI is not the right choice
Do not choose from vendor benchmark summaries alone. Compare Grok 4.7 with current Claude pricing and OpenAI pricing on your own prompts, tools, regions, retention requirements, and accepted-task cost. xAI's US regional endpoint carries a token premium, and tool-heavy workloads also add search, code-execution, file, or collection charges.
Price History
Only models with a recorded price change are charted here.
Price history tracking started April 2026. Flat model charts stay hidden until a price change is detected.
View pricing changelog →
Frequently asked questions
How much does xAI Grok cost per token?
For prompts below 200,000 input tokens, Grok Build 0.1 starts at $1 input / $0.2 cached / $2 output per 1M tokens, while Grok 4.7 is $2 input / $0.5 cached / $6 output. xAI applies the higher long-context tier once a request reaches 200,000 input tokens.
How much does Grok Bot cost?
xAI has not published a standalone Grok Bot price. Beta access is bundled with SuperGrok Heavy, Cursor Ultra, and Cursor Teams Premium. Subscriptions include weekly usage, and eligible accounts can add on-demand usage billed from the model and token cost.
How much does Grok Imagine Image 2.0 cost?
Grok Imagine Image 2.0 costs $0.04 for 1K low, $0.06 for 2K low or 1K medium, and $0.08 for 2K medium. Editing also adds $0.01 per input image. Auto quality currently maps generation to low and editing to medium.
When is grok-imagine-image-quality retiring?
xAI retires grok-imagine-image-quality on November 2, 2026. The slug will continue to resolve, but xAI will serve Grok Imagine Image 2.0 at low quality and bill the lower replacement rate.
How does Grok compare to GPT-5?
Price depends on the exact model and context length. Use the live comparison table rather than treating Grok Bot as a model: the Bot product chooses its serving model automatically and can fail over between models.
What models does xAI offer?
xAI offers Grok 4.7 as its current flagship, alongside Grok 4.6, Grok 4.5, Grok 4.3, dated Grok 4.20 reasoning, non-reasoning, and multi-agent SKUs, plus Grok Build 0.1. Older rows remain useful for migration comparisons.
What's the current Grok context window?
xAI lists 500K context for Grok 4.7, Grok 4.6, and Grok 4.5; 1M for Grok 4.3 and the dated Grok 4.20 SKUs; and 256K for Grok Build 0.1. Long-context pricing starts at 200K input tokens.
Can I access Grok without X Premium?
Yes. The xAI developer API and its per-token billing are separate from consumer Grok subscriptions. Grok Bot is also a separate beta product with plan-gated access.
Methodology
Pricing sourced from xAI's official developer pricing page on . All text rates are USD per 1 million tokens; long-context rates begin at 200K input tokens. We track 10 current and legacy Grok text rows from the daily pricing pipeline.
Compare xAI to other providers
Further reading: Full AI API pricing comparison · Cheapest AI APIs in 2026.