Elon Musk didn’t disappoint! SpaceX released the Grok 4.6 model as scheduled in early August, with upgrades focusing on long-running AI agents, complex programming tasks, and interactive and visual project development. According to their own published test results, Grok 4.6 scores on par with OpenAI’s GPT 5.6 Sol Max in multiple metrics, slightly below Claude Fable 5 Max. Simply put, Grok 4.6 is better at turning a somewhat vague product idea directly into a functional prototype application. It can independently research unfamiliar fields, build application frameworks, implement core features, and iteratively refine based on user feedback. When handling long-duration tasks, it also conducts more checks and tests on its outputs. Currently, the API pricing is $2 per million input tokens and $6 per million output tokens, while the faster version costs double. However, during the first week after launch, Cursor and Grok Build will offer double the usage quota within their plans.