Stories
SpaceX AI Unveils Grok 4.7: Faster, Cheaper, and More Powerful Coding Model
- •SpaceX AI launched Grok 4.7, its most capable model for coding and knowledge work, featuring robust safeguards.
- •Grok 4.7 excels in longer coding tasks, verifying its own work, and managing extended contexts, outperforming previous versions.
- •The model demonstrates enhanced capabilities in creating documents and presentations, as proven by its GDP Val benchmark performance.
- •Grok 4.7 incorporates an entirely new safeguard stack, offering strong refusal and jailbreak resistance, particularly in cybersecurity.
- •Priced at $2 per million input tokens and $6 per million output tokens, Grok 4.7 is available via Cursor, Grok Build, and the Grok API.
Key Questions Answered
Grok 4.7 is available now through Cursor, Grok Build, the Grok API, third-party coding harnesses, model routers, and cloud platforms. It is priced at $2 per million input tokens and $6 per million output tokens, with an additional faster variant offering twice the output speed at a higher price.
Grok 4.7 features an expanded base model and longer reinforcement learning runs focused on complex, multi-hour tasks. It excels at verifying its own work, managing longer contexts, natively understanding the Grok Bot harness for conversational tasks, and curating documents and presentations.
Grok 4.7 features an entirely new safeguard stack with strong resistance to refusals and jailbreaks. It achieves high utility for benign tasks while minimizing false refusals, restricting only 3.3% of potentially dangerous dual-use prompts on the HackerBench v0.3 benchmark.
