# Grok 4
Grok 4 is the fourth-generation [[Grok]] model from [[xAI]], the umbrella for the Grok 4.x minor releases including [[Grok 4.3]]. It is xAI's frontier multimodal [[Large Language Models (LLMs)|LLM]] line as of 2026, succeeding Grok 3.
## Positioning
- Closed-weight; hosted via X (`@grok`), grok.com, and the xAI API
- Multimodal across text and image; voice transcription is the strongest standalone capability in the line
- Headline competitor to [[Claude Opus 4.7]], [[GPT-5.4]], and [[Gemini 3]]
- Distribution moat: native integration with the X platform — tweet/thread analysis as a first-class workflow
## Variants
- [[Grok 4.3]] — May 2026 minor release; voice transcription accuracy, tone matching, token-density
Other minor versions and specialty variants (fast / mini / reasoning tiers) follow xAI's evolving tiering convention; check https://docs.x.ai/ for the current list.
## Why it matters
- Confirms xAI is shipping at frontier cadence; Grok 4 closed the gap with the established frontier labs
- Voice transcription and X-distribution form the differentiated moat rather than raw reasoning depth
- Keeps the 2026 LLM race a four-way contest (xAI, [[Anthropic]], [[OpenAI]], [[Google DeepMind]]) rather than two- or three-way
## Caveats
- xAI has not published full model cards at the 4.x minor-version granularity for every variant
- Comparison numbers come from community evaluations and competing-lab launches, not xAI's own benchmarks
- Tiering, pricing, and context windows change faster than the version numbers; verify against the live docs
## References
- xAI: https://x.ai/
- Grok docs: https://docs.x.ai/
## Related
- [[Grok]]
- [[Grok 4.3]]
- [[xAI]]
- [[Elon Musk]]
- [[Large Language Models (LLMs)]]
- [[AI Frontier Model]]
- [[AI Foundation Models]]
- [[Claude Opus 4.7]]
- [[GPT-5.4]]
- [[Gemini 3]]