The only AI with a live feed of X's 500 million daily posts — plus frontier reasoning, image and video generation, voice cloning, and a fast-growing work layer. We teach the capability AND the judgment, because nobody selling it will.
Models, DeepSearch, Imagine images + video, voice cloning, Connectors, Agent Mode — and the five-tier pricing ladder decoded, including why most people shouldn't buy Heavy.
Start here — real value before you ever upgrade.
The real-time X superpower and the class of work it serves, "truth-seeking" honestly assessed, free-tier reality, and three starter workflows with evidence-grading built in.
The tier maze (SuperGrok, Heavy, Premium+), the very cheap API, and the "free credits" data trade to watch out for.
SpaceXAI's strongest model, launched July 8 — frontier-adjacent coding and agents at $2/$6 per million tokens and 80 TPS. What their own charts show, where it runs today, and the EU caveat.
SpaceX bought xAI in February; July made the rebrand public. What changes for you (almost nothing), the naming nuance, and the summer ship list worth knowing.
The everyday workflows — included in both paid tiers.
Grok 4 vs 4.20 Reasoning vs 4.3, the one sorting question that survives every release, when Think mode earns its wait, and what Big Brain is for.
The research loop and how to steer it: the brief pattern, decision dossiers, claim autopsies — and the fair scorecard against Perplexity and ChatGPT research.
The lesson only Grok can star in: brand monitoring, competitor sweeps, trend triage, breaking-news verification ladders — with the bias caveats that keep it honest.
The five-slot prompt pattern, single-variable iteration, the image-to-image workhorse loop — and the three-question judgment layer before anything ships.
Drafting with the spice dialed deliberately, document interrogation with the Grok-only freshness pass, decision briefs — and the honest map of which daily jobs Grok wins.
The data dials to set today, Grok's distinctive failure mode, the companions sideshow in exactly one paragraph, and the clear-eyed when-NOT-to-use-Grok list.
Install the free Grok Office add-in, what each app's agent does, the live web/X edge — and the honest data trade-off vs. Microsoft 365 Copilot before you point it at company files.
Two AI sidebars, the same three apps. The neutral, honest breakdown of which to reach for — decided by where your information lives and how sensitive it is.
A Grok coding model arrived in GitHub Copilot and was pulled within a year. The honest case study in why you never hard-wire your workflow to a single AI model.
Video, voice, connectors, and the API — exclusive to Pro+.
10-second clips with native audio, the still→motion ladder, directing the infinite canvas like a shot list — and the deepfake line, drawn clearly.
Where voice genuinely wins, tier minute-meters, legitimate cloning uses — and the family code word that defeats voice-clone scams.
Email, calendar, docs, repos, MCP; Canvas in passes; Agent Mode briefed like a delegation — and the honest placement against Copilot and ChatGPT's work layers.
The cheap test of the top model, the $175/month free-credits data trade spelled out plainly, and the agentic coding model placed honestly against Claude Code and GitHub Copilot.
All 17 Grok lessons are available today — plus Copilot, Claude, ChatGPT, Gemini, Perplexity, OpenClaw, the hands-on builds, and the playground. Founding pricing (vs $19 standard) locks in for life.
Start free →