Example apps, recipes, and guides for building with the Grok API. Each one has its full code on GitHub.
Turn any article into audio that highlights each word as it's spoken, using the per-character timestamps from text to speech, and click a word to jump there.
Speak one language and hear Grok say it in another a moment later, from a page that connects straight to the realtime voice API with a short-lived token.
Make a drawing, a mascot, or a pet photo talk in a voice you pick, and have a second character answer, with Grok Imagine reference-to-video.
Serve a SQLite database as a small MCP server that only reads, and watch Grok list its tables, read their schemas, and run SQL through it to answer a question.
Give Grok 203 tools from a made-up company's eight work apps, show it only their names, and let it load the few each question needs with tool search.
Play one 30-message chat three ways to see what the prompt cache saves, how the time at the top of the system prompt breaks it, and what compaction costs.
Watch poisoned pages try to make an agent leak private data, see each of four defenses miss an attack another catches, and stop all ten with all four on.
Build a support agent that stops before a refund, waits for a person to approve it, and resumes the same run from a stored response, even after a restart.
Turn an article or a PDF into a two-host podcast episode, and watch it being written and recorded live in a small web app.
Turn a screenshot of a UI into a React component styled with Tailwind, then have Grok compare the result with the original and fix what doesn't match.
Prep for a sales call with a one-page brief, as Grok searches the web and ๐ and looks up the account through CRM functions that run in your app.
Drop in a CSV and ask questions about it. Grok writes and runs pandas in a sandbox, shows the code it ran, and answers with computed numbers and a chart.
Ask questions about a folder of documents and get answers that cite the passages they came from, with a metadata filter that decides which documents count.
Native iOS and Android apps for real-time voice conversations with Grok, plus text-to-speech on iOS.
A voice agent you can call on the phone, using Twilio.
Turn one photo into a time-lapse, like a street through the four seasons, by editing it for each stage and pinning every edit in one Grok Imagine video.
Turn one product photo into a vertical video ad with a voiceover, as Grok writes the brief, places the product in three scenes, picks one, and animates it.
Run a prompt's test cases at every reasoning effort, grade the answers with code checks and a judge, and find the cheapest effort that's still good enough.
Turn receipt photos and PDF invoices into a spreadsheet that adds up, with Grok reading each one twice and flagging the cells where the reads disagree.
Turn a thousand app reviews into ranked themes with counts, trends, and quotes by labeling every review in one Batch API job.
Turn a one-line premise into a four-shot short film with Grok Imagine keyframes, image-to-video, and narration, and watch it being made in a small web app.
Turn a few topics into a two-minute spoken briefing from ๐ and the web, with Grok skipping the stories it already told you about and a cost cap on every run.
A React client with swappable Python and Node.js backends that talk to the Voice Agent API over WebSocket.
Paste a link to a post on ๐, and Grok checks each claim on the web and on ๐ at the same time and writes a note where every sentence cites a source.
Score the sentiment about any topic from live ๐ posts with Grok's X Search tool, on a dashboard with a score for every post and the sentiment day by day.
Describe a bug in a small repo, and watch Grok reproduce it with a failing test, fix it, and rerun the tests, checking every command before it runs.
Caption a call live by speaker, keep a running list of decisions and action items, and ask questions whose answers link to the moment in the transcript.
Dub a video into another language, with each line timed to the original speech, shortened by Grok when it runs long, and mixed over the original background.
A low-latency browser voice agent that uses WebRTC between the browser and a server, which connects to xAI over WebSocket.