xAI releases its high-speed, high-performance inference model 'Grok Code Fast 1' for free for a limited time

xAI has revealed that it has built ' grok-code-fast-1, ' a high-performance coding engine, as a 'more agile and responsive solution optimized for everyday operations.'

Introducing Grok Code Fast 1, a speedy and economical reasoning model that excels at agentic coding.

Now available for free on GitHub Copilot, Cursor, Cline, Kilo Code, Roo Code, opencode, and Windsurf.

https://t.co/3tMbmLbxOP

— xAI (@xai) August 28, 2025

Grok Code Fast 1 | xAI
https://x.ai/news/grok-code-fast-1

grok-code-fast-1 is versatile across the software development stack, with particular expertise in TypeScript, Python, Java, Rust, C++, and Go. It is described as being able to complete common programming tasks with minimal supervision, such as building projects from scratch and providing insightful answers to code-based questions.

Grok-code-fast-1 was developed with a focus on ease of use and user satisfaction based on human evaluation, resulting in a model that performs well and is cost-effective for the tasks developers face every day.

xAI built grok-code-fast-1 from scratch, employing a completely new model architecture, carefully crafting a pre-training corpus rich in programming-related content, and then using a carefully curated, high-quality dataset for post-training that reflects real-world pull requests and coding tasks.

The price per million input tokens is $0.20 (approximately ¥30) via the xAI API, $1.50 (approximately ¥221) per million output tokens, and $0.02 (approximately ¥2) per million cached input tokens, demonstrating cost performance superior to other major models in terms of token count and output price per second. The table below shows the number of tokens per second on the vertical axis and the output price per million tokens on the horizontal axis.

Additionally, grok-code-fast-1 is available for free for a limited time from GitHub Copilot, Cursor, Cline, Roo Code, Kilo Code, opencode, and Windsurf. The deadline for free availability has not been announced at the time of writing.

xAI stated, 'Today's models are certainly powerful, but the inference and tool invocation loops can feel frustratingly slow and not designed specifically for proactive coding workflows. As heavy users of proactive coding tools, our engineers saw the need for a more agile and responsive solution optimized for their daily work. We look forward to rapidly iterating on this new model family based on your feedback.'