Grok 4.5 enters private coding beta

// 1h agoMODEL RELEASE

Grok 4.5 enters private coding beta

During Elon Musk's birthday, new details were shared regarding xAI's upcoming Grok 4.5 model, which has entered private beta testing. The model was trained with data from Cursor and is expected to deliver a substantial leap in agentic coding and broader capabilities.

// ANALYSIS

Training a foundation model on Cursor's developer interaction data signals a strategic shift towards acquiring high-intent agentic coding telemetry to beat competitors on code-generation tasks.

–Developer telemetry: Leveraging Cursor data could significantly improve the model's understanding of real-world developer workflows and iterative debugging.
–Private beta phase: Restricting access allows xAI to refine safety boundaries and agentic tool-use reliability before a wider release.
–Competitive landscape: Focus on agentic coding positions Grok 4.5 to directly challenge established code-generation models like Claude 3.5 Sonnet.

// TAGS

grok-4.5xaicursoragentic-codingllmartificial-intelligenceprivate-beta

DISCOVERED

1h ago

2026-06-28

PUBLISHED

2h ago

2026-06-28

RELEVANCE

8/ 10

AUTHOR

mark_k

// KEEP READING

More AI developer news from the feed

EXPLORE FULL FEED

UPDATE2h ago

Grok Build 0.2.70 lands terminal improvements

xAI has launched Grok Build version 0.2.70, focusing on developer terminal and agentic workflow enhancements. Key updates include the introduction of the grok wrap command wrapper for local clipboard support, alongside a keyboard shortcut (Ctrl+4) to toggle the prompt queue inside macOS VS Code terminals.

NEWS3h ago

Levelsio codes on VPS using Claude Code

Pieter Levels (levelsio) shared that he has coded almost exclusively on a VPS using Claude Code for nearly a year. He praised the setup for its multi-device flexibility and low battery impact, noting that the AI tool caused site issues only twice during this period.

UPDATE5h ago

Paperclip integrates Hermes for zero-human companies

Paperclip's v2026.626.0 release integrates Nous Research's Hermes agent natively, bringing tool-use and persistent memory directly into the platform. This update enables developers to build and manage autonomous zero-human companies using a visual orchestration dashboard.