SpaceXAI's Grok 4.6 scored 61 on the Artificial Analysis Intelligence Index, placing it alongside GPT-5.6 Sol at the intelligence frontier, according to artificialanalysis.ai. Released just over a month ago, Grok 4.6 improved by 5 points over its predecessor, Grok 4.5, and 23 points compared to Grok 4.3, marking a significant leap in AI performance.
The model demonstrated strong agentic capabilities, achieving a GDPval-AA v2 Elo rating of 1753, ranking just behind Claude Opus 5 and closely matching Claude Fable 5 and Qwen3.8 Max. Grok 4.6 scored 50.7% on ð³-Banking, placing it among the top two models alongside Qwen3.8 Max, and achieved 88.4% on Terminal-Bench v2.1, aligning with leading AI models. Pricing remains unchanged from Grok 4.5, at $2/$6 per million input/output tokens, significantly lower than competitors Claude Opus 5 and GPT-5.6 Sol.
This performance positions Grok 4.6 at the forefront of AI intelligence while maintaining cost efficiency, with a cost per task of $0.84, comparable to Kimi K3 but with slightly higher intelligence. The model's balance of high intelligence and lower cost places it on the Intelligence versus Cost per Task Pareto frontier, highlighting SpaceXAI's competitive edge in the AI market dominated by Anthropic and OpenAI models.
The Artificial Analysis Intelligence Index update on August 12, 2026, confirms Grok 4.6's status as a leading AI model, with headline pricing and performance metrics publicly available on artificialanalysis.ai.