SpaceXAI Launches Grok 4.7 for Enhanced Coding and Complex Tasks
- Grok 4.7 scored 46.3% on CursorBench 4.0, surpassing Grok 4.6’s score of 40.4%.
- The model is priced at $2 per million input tokens and $6 per million output tokens.
- In DeepSWE v1.1, Grok achieved a score of 71%, slightly below GPT-5.6 Sol’s score of 72.7%.
- Grok scored only 38% in terminal-based tasks, improving from Grok 4.6’s score of just over half that amount.
- Cybersecurity tests revealed Grok allowed for a dangerous prompt rate of only about 3.3% in HackerBench v0.3.
The release of Grok 4.7 marks a significant upgrade with improved performance metrics across various benchmarks, although it still trails behind competitors like Fable 5.1 in some areas.
Despite its competitive pricing, Grok 4.7’s performance varies significantly by task type, as seen with its scores across different benchmarks such as the Intelligence Index and terminal-based tasks.(Source)
