Model Details
Grok 4.7
Quick answer: Grok 4.7 is xAI's coding- and agentic-work-focused model, released September 21, 2026. xAI pitches it as twice as fast and half the price of comparable models versus Grok 4.6, with gains on agentic coding (Terminal-Bench 4.0, DeepSWE), long-horizon office work (AA Briefcase), and real-world agent Elo (GDPVal-AA).
What Grok 4.7 Is
Grok 4.7 is xAI's successor to Grok 4.6, aimed at coding and knowledge-work agents. It ships with a standard tier and a faster variant (2x speed at 2x price), keeping the same $2/$6 per-million-token base pricing as Grok 4.6 for the standard tier.
Specifications
| Organization | xAI |
| License | Proprietary (API only) |
| Release date | September 21, 2026 |
| Modality | Text |
| Knowledge cutoff | Not disclosed |
Public Benchmark Scores
| Benchmark | Score | Source | Date |
|---|---|---|---|
| DeepSWE v1.1 | 71.0% (high effort) | xAI | Sep 2026 |
| AA Briefcase v1.1 | 1,657 | xAI | Sep 2026 |
| Terminal-Bench 4.0 | 37.6% (xhigh effort) | xAI | Sep 2026 |
| GDPVal-AA v2 | 1,695 Elo (xhigh effort) | xAI | Sep 2026 |
| Harvey Legal Agent Benchmark | 19.6% (xhigh effort) | xAI | Sep 2026 |
Scores as reported in xAI's own Grok 4.7 announcement (x.ai/news/grok-4-7, September 2026); not independently verified by Benchgen.