Skip to content
NewsProduct launch

xAI ships Grok Build 1.0.0, moving the coding CLI agent from beta to stable

· by Pondero Newsdesk

The short version

xAI released Grok Build 1.0.0 on August 7, 2026, the first stable-versioned release of its coding CLI agent, shipping fixes for MCP screenshot corruption, plan-review interactivity, and session reliability.

xAI ships Grok Build 1.0.0, moving the coding CLI agent from beta to stable

xAI jumped Grok Build from 0.2.x beta versioning to a stable 1.0.0 release on August 7, 2026 per the official xAI changelog, shipping fixes for MCP screenshot corruption, plan-review interactivity, and session reliability in the same push. This is the first stable-versioned release of the tool.

What

The 1.0.0 release contains more than 30 individual changes. The most operationally significant is the MCP screenshot fix: tools that return images previously dropped or corrupted large screenshots, a silent failure that affected any agentic pipeline routing visual output through an MCP server. That issue is closed in 1.0.0.

Interactive use during plan review also works for the first time. The model picker and command palette were locked while a plan was visible, which forced users to approve or reject before adjusting settings or switching models. That constraint is gone in 1.0.0.

Dashboard orientation improved as well. Each session row now displays a one-line summary of the agent's last turn, replacing the blank row that required opening a session to know where it left off. The extensions modal now sorts items alphabetically with collapsible Skills sections, replacing a flat unsorted list.

Terminal compatibility received fixes for two common remote-work setups: auto theme detection works over SSH and inside tmux. Large repositories with many deny-glob entries now start the sandboxed agent without hanging. Session integrity tightened on two edges: Esc and stop now prevent background tasks from resuming the model after a cancel, and rapid message queuing no longer drops earlier entries.

Codebase restore on large or shallow git repositories no longer hangs. Remote resume defaults to restoring conversation history only; code restoration must be explicitly opted in via a command-line flag.

Why it matters

The 1.0.0 designation is a signal, not just a number. A stable semver tag is the threshold at which version-pinning and dependency tracking become practical for teams running the tool in CI or shared dev environments. Shops evaluating Grok Build alongside Claude Code, Codex CLI, and Gemini CLI now have a stable release to compare against, rather than a moving beta.

The MCP screenshot fix closes a real data-loss path. Any workflow that passes image output from an MCP tool to the agent was silently dropping content before 1.0.0. Teams that hit this and worked around it with text extraction can reconsider.

The plan-review interactivity fix is smaller but affects the daily loop: users who kept the model picker open as a reference while reading a plan can do that without losing state.

What to watch next

xAI has not announced consumer pricing or expanded API access for Grok Build alongside the 1.0.0 release. The tool currently requires an xAI API key with no announced free tier. Whether a lower-cost or open-access tier follows the stable designation is the next practical threshold for teams sitting on the sideline.

Sources