xAI Launches Grok 4.7. It's Bigger, But Late to the AI Frontier Party
xAI released Grok 4.7 as its newest flagship model, after multiple delays, and says it is faster and cheaper than Grok 4.6 while spending more time on difficult reasoning and using stronger safety guardrails. It is available immediately in the Grok app, Cursor, Grok Build, and xAI’s API, with pricing at $2 per million input tokens and $6 per million output tokens. The model has 2.1 trillion parameters, up from 1.5 trillion, and includes extra training data from SpaceX sources such as Starlink telemetry, manufacturing records, and failure logs to improve performance on hardware and physical systems. Benchmark results show improvement but not a frontier lead: Grok 4.7 trails Claude Fable 5.1 on GDPval, AA-Briefcase, and CursorBench. The launch reinforces xAI’s strategy of competing on cost and availability rather than outright best-in-class capability. Musk also said future versions, including Grok 4.8, 4.9, and 5, should deliver bigger jumps.
