ChatImageVideo

Model developer guide

xAI

14 website model guides

Use Grok for a written answer and Grok Imagine for visual output. Choosing between chat, an image edit and a moving shot matters more than the shared brand name. This guide pairs checkable creative briefs with the distinct model IDs, request types and app boundaries behind them.

ResearchSpeedCreative generation

Practical guidance

Where should I start?

Editorial starting points, not a performance ranking. Check current app access before choosing a model.

Review a difficult written or coding problem

Evaluate Grok 4.7 for reasoning over supplied text and images. The official model reference supports function calling and structured outputs, useful integration capabilities when a host actually exposes them.

Tradeoff: Reasoning and a larger context do not prove an answer is correct. Ask for source locations and test proposed code rather than relying on confident explanations.

Grok 4.7 guide

Generate a visual concept or controlled revision

The image API provides text-to-image generation with aspect-ratio and resolution parameters. Treat editing as its own workflow, with a source image and explicit preservation instructions where app support is confirmed.

Tradeoff: A family-level API feature may apply only to a particular model revision. New quality controls in current documentation are not automatically controls for every catalog entry.

Grok Imagine guide

Develop a short, single-action shot

xAI documents asynchronous video generation from text or a starting image, with separate reference and editing workflows. Plan one subject action and one camera move before adding complexity.

Tradeoff: Grok Imagine Video 1.5 supports native 1080p and clips up to 15 seconds, with sound and up to 14 reference images.

Grok Imagine Video 1.5 guide

Technical details

What does the provider document?

Official documentation describes the named model or gateway—not every feature available in this app.

Grok 4.7 has a specific reasoning contract

The grok-4.7 reference lists a 500,000-token context, text/image input and text output. Supported reasoning efforts are low, medium, high and xhigh, with high as default. It supports function calling and structured outputs but does not list Batch API support.

Image controls and result delivery

xAI's image generation API documents aspect_ratio, resolution and response_format, with URL or base64 output options. Current documentation limits its quality parameter to grok-imagine-image-2.0; that is not a universal feature of older Imagine models. Hosted result URLs are temporary, so app storage must be checked independently.

Video generation, references and edits are distinct

The video API returns a request ID before a result is ready. Generation supports different text, image and reference modes; editing uses its own endpoint and constraints. Image-to-video follows the input image's aspect ratio. Current 1.5 documentation adds frame and voice features, but this is provider capability evidence, not proof of app passthrough.

Complete website coverage

Explore all 14 model guides

Every existing catalog entry for this provider is linked below. Status labels come from the individual website guides; catalog coverage is not a guarantee of current app access. Historical and snapshot pages remain references, not extra active models.

Practical guidance

From a brief to a useful result

Original prompt examples, not generated-output demonstrations. They have not been generation-tested.

Workflow 1

Turn a support incident into a testable diagnosis

Provide redacted logs, observed behavior and a small relevant code excerpt. State whether the model has any tools; do not ask it to pretend to execute a test.

Analyze this incident using only the logs and code supplied. Build a short timeline, distinguish observations from hypotheses, and rank three possible causes by supporting evidence. For each cause propose one discriminating test and its expected result. Cite the relevant log entry or function. Do not claim a fix or a test succeeded without evidence.

Check the result: Confirm the cited log entries exist, timestamps align and each proposed test could distinguish the competing causes.

Read the Grok 4.7 guide

Workflow 2

Storyboard a calm reveal

Choose the audited 1.5 video option if available. Use the actual duration and resolution controls in the app; this prompt describes motion rather than setting numerical API parameters.

Create one continuous shot of an original ceramic teapot on a pale wooden table beside a rain-streaked window. Begin with the teapot centered and slowly move the camera a small distance to the right. Keep its shape, handle and spout consistent. Let soft window reflections shift naturally. No cuts, added objects, lettering or sudden movements.

Check the result: Watch the handle and spout throughout the clip, check the requested camera direction and reject object morphing. Inspect the actual duration and dimensions separately.

Read the Grok Imagine Video 1.5 guide

Technical details

What changes inside EZ Ai Assist?

Official xAI documentation was reviewed for the capabilities cited here. Existing image and video owner audits describe app behavior, not independently observed generation results. Chat request settings and new provider features were not assumed to be enabled in EZ Ai Assist.

Owner image audit · October 8

Image inputs in the app

The existing guides report direct xAI image routing and use of only the first attached reference for editing. This does not establish batch requests, every provider quality option or all current reference-image capabilities in the app.

Owner video audit

Video aliases and narrower controls

The Video 2/v2 aliases are reported to route to grok-imagine-video-1.5. That route supports an audited reference array and audio-off passthrough, while Lite has no audited reference-array or editing support. Its 1080p option is reported as upscaled from 720p; read the exact model guide for bounds.

Not live-tested

What is not verified

Final chat reasoning settings, tool execution, storage retention and new xAI keyframe or voice passthrough were not inspected. A provider documentation update cannot by itself establish a new app feature, a successful video request or the identity used by an old alias.

Before you rely on a result

Limits, costs and tradeoffs

  • A text model with image input does not generate images. Likewise, an edited still image and an image-anchored video are different request types with different constraints.
  • Larger resolution labels do not always mean native synthesis at that size. Distinguish the audited Lite upscale path from 1.5 generation, and assess the downloaded result.
  • Current documentation can describe newer models than the website catalog. It is not permission to silently replace existing identities, promise new controls or turn provider reference rates into customer charges.

Provider references and picker estimates are not EZ Ai Assist customer charges. Read the website pricing explanation.

Compare by task

Other paths to explore

Google

Compare evidence-based multimodal analysis or another video workflow, checking each route's input contract.

OpenAI

Evaluate text reasoning and image revision against the same test brief.

Runway

For image-anchored or editing-led video, compare the source-media requirements rather than only a version label.

Common questions

xAI, without the assumptions

Why does the Video 1.5 guide have a Video 2 address?

The existing URL is preserved. The owner audit maps the older Video 2/v2 app aliases to grok-imagine-video-1.5. The guide makes that identity explicit rather than inventing another provider model.

Does Grok automatically search the web for every answer?

No such app behavior is established here. Model tool support and a configured, executed search workflow are different. Ask for evidence and verify the tool configuration before relying on live information.

Can I use every xAI keyframe or voice feature here?

The provider's latest video documentation describes additional controls, but the app audit does not confirm all of them. Plan around the exact model guide's reported controls and label anything else unverified.

New to the terminology?

Official documentation

Sources & review notes

Highlighted provider contracts reviewed . Individual model guides carry their own review dates. Sources can change; these pages do not perform live health checks.

Official xAI documentation was reviewed for the capabilities cited here. Existing image and video owner audits describe app behavior, not independently observed generation results. Chat request settings and new provider features were not assumed to be enabled in EZ Ai Assist.