xAI launches Grok 4.6 with frontier-model benchmarks and unchanged pricing

xAI launches Grok 4.6 with frontier-model benchmarks and unchanged pricing

N
News Editor
2026-08-12 17:07:37
xAI has rolled out Grok 4.6, the latest version of its flagship model, with Elon Musk announcing the release on X on Aug. 12 and describing it as “smarter, faster, and insanely cost-effective.” The company’s official account called it a major upgrade over Grok 4.5 while keeping pricing unchanged. According to benchmark results published by xAI, Grok 4.6 scored 61 on the AA Intelligence Index, matching OpenAI’s GPT-5.6 Sol, and outperformed it on CursorBench, FrontierCode, and AA-Briefcase. Musk also said the model reached an ELO score of 1753. On pricing, xAI kept the Grok 4.5 rate of about $2 per million input tokens and $6 per million output tokens. The company said performance is roughly comparable to Fable 5 Max while costing about 15% as much, and said input and output pricing is about 80% and 88% lower than competing frontier models. xAI also said Grok 4.6 keeps the same 1.5 trillion-parameter V9 base model used in Grok 4.5, with gains coming mainly from post-training through supervised fine-tuning and reinforcement learning. The model is now available through the xAI API, Grok App, grok.com, and X Premium+ and SuperGrok subscriptions.

xAI has launched Grok 4.6, the newest version of its flagship model. Elon Musk said in a post on X on Aug. 12 that Grok 4.6 is “smarter, faster, and insanely cost-effective,” while xAI’s official account described it as a major upgrade from Grok 4.5 with no change in price.

Benchmark claims focus on higher performance at the same price

The release is centered on value. Based on benchmark data published by xAI, Grok 4.6 scored 61 on the AA Intelligence Index, matching OpenAI’s GPT-5.6 Sol. xAI also said the model outperformed GPT-5.6 Sol on CursorBench, FrontierCode, and AA-Briefcase. Musk added that the model posted an ELO score of 1753.

Pricing remains the same as Grok 4.5 at roughly $2 per million input tokens and $6 per million output tokens. xAI said Grok 4.6 delivers performance around the level of Fable 5 Max while costing about 15% as much. It also said pricing is about 80% lower for input tokens and about 88% lower for output tokens than rival frontier models. Those figures were published by xAI, and the article noted that third-party validation is still needed.

Base model stays the same, with gains coming from post-training

On the technical side, Grok 4.6 uses the same 1.5 trillion-parameter V9 base model as Grok 4.5. The main improvements come from post-training, including supervised fine-tuning, or SFT, and reinforcement learning, or RL.

In other words, xAI did not swap in a new foundation model for this release. The update instead focuses on training the same model more effectively for practical use.

Available through API, app, website, and X subscriptions

Grok 4.6 is now available through the xAI API, the Grok App, grok.com, and X’s Premium+ and SuperGrok subscription plans.

ABMedia said Chain News had previously reported that xAI had rolled out the “AI coworker” Grok Bot and had signaled that Grok 4.6 would arrive this week. The launch has now been completed. For developers, the article said xAI is still pressing OpenAI and Anthropic with a pricing strategy aimed at winning market share.

This article was originally published by Bit.Fan. For more cryptocurrency news and market insights, visit www.bit.fan.
430

Disclaimer:

The market information, project data, and third-party content displayed on this platform are for industry information sharing only and do not constitute any form of investment advice or return commitment.

Cryptocurrency trading carries high risks. Users should fully assess their risk tolerance and make independent decisions. All profits, losses, and legal responsibilities are borne by the users themselves.