Grok 4.6 scores 61 on the Artificial Analysis Intelligence Index

Grok 4.6 has achieved a remarkable score of 61 on the Artificial Analysis Intelligence Index, just a month after its release, marking its return to the forefront in intelligence alongside OpenAI. Positioned behind only Anthropic, Grok 4.6 excels in agentic performance, achieving an Elo of 1753 in GDPval-AA v2, a measure of real-world agentic tasks.

This score places it just behind Claude Opus 5 and in close competition with Claude Fable 5 and Qwen3.8 Max. Grok 4.6's pricing remains a standout feature, unchanged at $2/$6 per 1 million input/output tokens, offering a significant cost advantage over competitors. Despite its advanced capabilities, it maintains a context window of 500k tokens and costs $0.84 per task, comparable to Kimi K3 while offering superior intelligence.

The model's efficiency is particularly noteworthy, completing tasks in approximately half the turns and a quarter of the input tokens required by leading competitors like Claude Opus 5. This blend of high performance and cost efficiency positions Grok 4.6 uniquely on the cost vs. performance Pareto frontier for agentic evaluations, making it an attractive option for those seeking advanced intelligence at a favorable price point.