xAI Grok-1 at a glance
xAI published the raw Grok-1 base checkpoint, a 314B mixture-of-experts model with about 25% of weights active per token, under Apache 2.0.
What changed with xAI Grok-1
On March 17, 2024, xAI released the weights and architecture of Grok-1, a 314B-parameter mixture-of-experts model in which about 25% of the weights are active for a given token.
The release was the raw base model from pre-training that ended in October 2023, not the version fine-tuned for chat. Weights and code were published under Apache 2.0 on GitHub and Hugging Face.
xAI Grok-1 in numbers
Values are copied from the linked source on the date shown. They are not IA Compa measurements.
- Total parametersas of · Source for Total parameters: x.ai (opens in a new tab)
- 314B
Where to compare xAI Grok-1
Independent leaderboards and official resources for comparing this model yourself. Rankings change often; each site shows its own current data.
General leaderboards
Why xAI Grok-1 matters for businesses
For most businesses, Grok-1 was too large and too raw to use directly, since it needed multi-GPU servers and further tuning. Its significance was that a frontier-scale lab released a big model under a fully permissive license, adding pressure on others to open up.
Sources for this story
- Official
- grok-1 repository (opens in a new tab)Official
xAI on GitHub · github.com
- Grok-1 model page (opens in a new tab)Benchmark
Artificial Analysis · artificialanalysis.ai
Last verified: . IA Compa summarizes public announcements for business readers. We are not affiliated with the companies mentioned. Always confirm details, pricing, and availability with the official source.
- #llm
- #xai
- #grok
- #open-weights
- #apache-2-0
- #mixture-of-experts
