xAI Brings Out Grok 4 Fast as its Latest Cost-Efficient Reasoning Model Promising Frontier-Level Performance and High Token Efficiency

The new Grok 4 Fast that has now been brought out by xAI has been built on Grok 4’s learnings, and it notably features the best cost efficiency and the best search capability on web and X. Relying on its 2M token context window and the unified architecture with reasoning and non-reasoning modes, it is also capable of delivering peak performance too.

Check out more about its features and capabilities below.

xAI’s New Grok 4 Fast Introduced

The Grok 4 Fast AI model features a maximized intelligence density thanks to its large-scale reinforcement learning, and in terms of its performance-level, it has outperformed Grok 3 Mini (with reduced token costs) and has achieved a similar performance as that of the Grok 4 (with 40% fewer average thinking tokens). To note, the 40% reduced tokens combined with the lower pricing enables the new Grok 4 Fast to achieve 98% cost efficiency to deliver the same performance of the Grok 4 model – thereby exhibiting a state-of-the-art price-to-intelligence rate (when compared to other such models on Artificial Analysis Intelligence Index).

Speaking more, the Grok 4 Fast has also been trained with tool-use reinforcement learning, meaning it knows exactly when to make use of tools for web browsing and code execution. On LMArena’s Search Arena, the Grok 4 Fast has secured the first position, setting a new benchmark on the general domain too. Additionally, it also achieved the 8th position in the LMArena’s Text Arena, which portrays its intelligence density. As mentioned, another key feature of Grok 4 Fast is its unified architecture. With the reasoning and non-reasoning modes together being handled in a single model and controlled via prompts, both end-to-end latency and token costs are being reduced.

About its availability, xAI’s Grok 4 Fast is currently available for all users on grok.com, iOS, and Android. It will also be available on OpenRouter and Vercel AI Gateway for free, but only for a limited time. Regarding the pricing details on xAI API, for less than 128k tokens, it is priced at USD 0.20 per 1M input tokens/USD 0.50 per 1M output tokens/USD 0.05 per 1M cached input tokens and for 128k or more tokens, it is priced at USD 0.40 per 1M input tokens/USD 1.00 per 1M output tokens/USD 0.05 per 1M cached input tokens.

Also to add, the brand invites feedback and based on which, plans to further improve its Grok 4 Fast model.