xAI has officially launched Grok 4.7, a new large language model optimized for programming and complex knowledge-based tasks. The model features enhanced self-verification capabilities, improved long-context management, and native support for the Grok Bot framework, allowing it to function as a persistent intelligent agent.
The model demonstrates significant performance gains in various benchmarks. In CursorBench 4.0, it scored 46.3% compared to 40.4% for the previous version. It achieved 71% in DeepSWE, 64% in EEBench, 19.6% in the Harvey Legal Agent Benchmark, and 56.7% in HealthBench Professional.
xAI has maintained the pricing structure of its predecessor, with input costs at $2 per million tokens and output costs at $6 per million tokens. The company positions this as a competitive alternative, offering performance comparable to industry leaders at roughly half the price of similar models.
The new release includes a rebuilt security architecture designed to improve defense against jailbreak attempts and malicious requests. The model is currently available to users via API and the Grok platform for applications ranging from software development to automated document creation.
This launch marks the third major update from xAI in less than three months, following the releases of Grok 4.5 in July and Grok 4.6 in August. The rapid iteration cycle underscores the company's strategy to challenge market leaders through aggressive pricing and accelerated product development.
