xAI Launches Grok 4.20 Beta: The First Grok That Learns After Deployment
The First Grok That Learns After Deployment
xAI launched Grok 4.20 as a public beta on February 17, introducing what the company calls its most breakthrough version of the series. The defining feature is a rapid learning architecture: unlike previous Grok models, Grok 4.20 continues to update its capabilities after deployment by incorporating user feedback on a weekly cadence. xAI release notes position this as the beginning of a new paradigm in continuously-evolving AI models.
Four-Agent Collaboration System
Grok 4.20 introduces a multi-agent system in which four specialized agents work simultaneously to tackle complex problems from different angles. The release also adds medical document analysis via photo upload — users can photograph a prescription or lab report for an AI analysis — alongside improved engineering reasoning capabilities.
Tesla Integration and Availability
Grok AI has also arrived in Tesla vehicles in Europe via the 2026.2.6 software update, enabling the AI to add and edit navigation destinations and act as an in-vehicle personal assistant. Grok 4.20 beta is available on x.ai.
Related Articles
HN treated Grok Build less as a feature drop and more as a test of control: an open Rust coding agent is useful, but telemetry, forks, and provider lock-in shaped the discussion.
xAI has released Grok 4.3 on its API, claiming top spots on agentic tool calling and instruction-following leaderboards, and ranking #1 in enterprise domains such as case law and corporate finance. It supports a 1M token context window at $1.25/M input and $2.50/M output.
xAI’s next Grok foundation model is moving from training into fine-tuning at 1.5T parameters, three times the size of the current 0.5T production model. Musk says Cursor data was added and public release is 2 to 3 weeks away.