Grok 4.7 ครึ่งราคาคู่แข่งระดับเดียวกัน Terminal-Bench พุ่งเกือบเท่าตัว

Grok 4.7, an AI model from SpaceXAI, has just been released at a price nearly equal to its competitors at the Terminal-Bench level, according to Nokka (นก-กา). This development comes amidst a week-long price war in the AI model market. Grok 4.7 was launched on November 21, priced at $2 per million tokens for input and $6 for output, following an extensive training period using reinforcement learning.

The model demonstrates significant improvements in handling long prompts and managing extended contexts, with performance gains seen in Terminal-Bench and Harvey Legal Agent benchmarks. Notably, Grok 4.7 excels in cost-efficiency, offering a price-to-performance ratio more favorable than competitors like GPT-5.6 Sol Max and Fable 5.1 Max.

The model also excels in cybersecurity and health domain tasks, achieving high scores on biosafety and HackerBench v0.3 benchmarks. Grok 4.7 is currently available via Cursor, Grok Build, Grok API, coding harnesses, model routers, and cloud platforms. For organizations considering adopting this AI model, it is recommended to test its performance on real-world tasks before making a decision.