Grok Build adds Grok 4.7 reasoning levels
xAI has integrated Grok 4.7 into its coding agent Grok Build, allowing developers to choose between Low, Medium, High, and Extra High reasoning effort levels depending on task difficulty. The upgrade brings substantial performance improvements over Grok 4.6, highlighted by a leap from 40.4% to 46.3% on CursorBench 4.0 and a near-doubling from 20.3% to 38.0% on Terminal-Bench 4.0.
Nearly doubling its Terminal-Bench score proves xAI is taking autonomous command-line developer agents seriously and positioning Grok Build as a direct challenger to tools like Claude Code.
• Selectable reasoning levels (Low to Extra High) give developers control over latency, cost, and deep problem-solving for routine tasks versus complex debugging.
• The dramatic improvement on Terminal-Bench 4.0 (20.3% to 38.0%) indicates superior environment navigation, multi-step shell command execution, and autonomous error recovery.
• The CursorBench 4.0 increase to 46.3% confirms stronger codebase context handling and multi-file code editing accuracy.
DISCOVERED
1h ago
2026-09-21
PUBLISHED
1h ago
2026-09-21
RELEVANCE
AUTHOR
XFreeze