What is Grok
Grok is strongest when users want a live, internet-connected assistant with a more consumer-facing personality and direct ties into the xAI and X ecosystem. It combines chat, search, voice, image generation, and API access in a way that feels broader than a single chatbot product.
Grok emerged from xAI’s effort to build a real-time, internet-connected assistant closely linked to X and a fast-moving media ecosystem. Since launch it has expanded from a chat product into a broader stack that now includes search, media generation, voice features, and APIs, which makes its trajectory important for anyone tracking consumer AI platforms that are becoming developer platforms too.
Core offerings
- Consumer chat with real-time web and X search.
- Image and video generation through the Imagine family.
- Voice features and developer-facing APIs.
- Multiple paid tiers spanning individuals, teams, business, and enterprise.
Pricing
As of July 18, 2026, xAI’s public pricing page lists a Free plan at US$0 per month and SuperGrok at US$30 per month. The Free plan includes real-time web and X search, voice mode, SOC 2 Type I and II compliance, and connectors. SuperGrok adds access to the Grok 4.5 model, connectors, higher rate limits across features, Expert mode, SOC 2 Type I and II compliance, and image and video generation.
xAI’s Business page separately lists Grok Business at US$30 per user per month for small-to-medium teams. The Business product page and license-management docs describe team seat management, consolidated billing, role-based access control, domain verification and claims, user analytics, custom data retention, connectors, Grok Build, and increased rate limits as part of the business workspace offering. Enterprise is contact-sales only and adds custom SSO, SCIM, custom RBAC, advanced user and access management, dedicated onboarding, customer-managed encryption keys, application-level encryption, and a dedicated data plane.
On the API side, xAI’s developer pricing docs currently list grok-4.5 at US$2.00 per million short-context input tokens, US$0.50 cached input, and US$6.00 output, with higher long-context pricing. xAI also publishes separate pricing for image, video, and voice APIs.
Model footprint
xAI publicly lists Grok 4.5, Grok Build, Grok 4.20 variants, plus media generation and voice endpoints. That gives buyers a more diverse product set than Grok’s consumer branding first suggests.
Why select Grok
Grok is relevant for teams that want current-event awareness, a consumer-friendly interface, and a vendor whose product is expanding beyond chat into media and developer tooling. It is also worth watching because xAI is moving quickly across both product and API surfaces.
Official sources: Grok pricing, Grok Business, xAI license management, xAI developer pricing, xAI models.
Current models
xAI publishes a relatively developer-friendly current model table, including context windows and short versus long context rates for the text API.
| Model | Context / output | Knowledge / training | Pricing / token use | Speed / notes |
|---|---|---|---|---|
Grok 4.5grok-4.5 | 500K context | Knowledge cutoff: not stated on the model page Training data: not publicly disclosed | Short context: US$2 input / US$0.50 cached / US$6 output per 1M tokens Long context: US$4 input / US$1 cached / US$12 output | Frontier Grok model with reasoning, structured outputs, and function calling |
Grok Build 0.1grok-build-0.1 | 256K context | Knowledge cutoff: not publicly disclosed Training data: not publicly disclosed | Short context: US$1 input / US$0.20 cached / US$2 output per 1M tokens Long context: US$2 input / US$0.40 cached / US$4 output | Lower-cost build-oriented tier for developer and agent workflows |
Grok 4.3grok-4.3 | 1M context | Knowledge cutoff: not publicly disclosed Training data: not publicly disclosed | Short context: US$1.25 input / US$0.20 cached / US$2.50 output per 1M tokens Long context: US$2.50 input / US$0.40 cached / US$5 output | Broader-context Grok tier still live in the API price book |
| Grok 4.20 reasoning / non-reasoning / multi-agent | 1M context | Knowledge cutoff: not publicly disclosed Training data: not publicly disclosed | US$1.25 input / US$0.20 cached / US$2.50 output per 1M short-context tokens | xAI currently exposes specialised 4.20 variants for reasoning and multi-agent behaviour |