Key highlights –
- xAI launched Grok Code Fast 1 on Thursday, aimed to prioritize speed over capability in coding specific workflows.
- The AI platform has partnerships with Copilot, Cursor, and Windsurf and more.
- Comes with free access via GitHub Copilot, until September 2nd.
On August 28, Elon Musk’s AI company, xAI launched Grok Code Fast 1, a dedicated agentic coding model “built from scratch, with a brand new architecture.”
We’re thrilled to introduce grok-code-fast-1, a speedy and economical reasoning model that excels at agentic coding.
Interestingly, Musk’s AI team has collaborated with the likes of GitHub Copilot, Cursor, Cline, Roo Code, Kilo Code, opencode and Windsurf to offer the coding model for free for a limited time.
In Grok Code Fast 1 introduction blog, xAI shared that the agentic coding model is “speedy and economical.”
You may also like to read: Grok 2.5 and Grok 3 to Become Open-Source Models: Elon Musk
What Can Grok Code Fast 1 Do?
The newly launched agentic model is versatile in coding languages including TypeScript, Python, Java, Rust, C++ and Go. The AI company claims that the model can complete common programming tasks with “minimal oversight.”
Musk’s company has also rolled out a prompt engineering guide for Grok Code Fast 1 for ease of use for its users. This caters to both developers using agentic coding tools as well as devs building coding agents via xAI API.
The general theme of the model’s working is – that fast enough looped would beat a slow yet theoretically accurate results. xAI defends itself for not being the best in accuracy, claiming that what truly matters in agentic coding is quick tool loops, steerability, and stable hand-offs. This is important as rivals will now have to step up on the speed parameter with the launch of Grok Code Fast 1.
The Grok Code Fast 1 official introduction blog also promised that consistent updates will be delivered for Code Fast 1, and users could expect it within days, instead of a weekly update timeline.
How Does The Model Help Devs
Musk’s latest agentic model comes with a free trial period until September 2. All interested developers can get a hands on access to it via GitHub Copilot, Cursor and more. Many of the Grok Code Fast 1 launch partners joined in to share their testimonials. Here are some excerpts –
Grok Code is seriously fast! We released the model codenamed as “sonic” in Cursor, and developers were very impressed by its speed.
– Lee Robinson
(VP of Developer Experience, Cursor
GitHub’s Chief Product Officer too had promising things to say for the model –
In early testing, Grok Code Fast has shown both its speed and quality in agentic coding tasks. Empowering developers with powerful tools is a core part of our mission at GitHub Copilot, and this is a compelling new option for our developers.
The model is generally available via the xAI API, priced at $0.20 with 1M input tokens, $1.50 for 1M output tokens, and $0.02 for 1M cached input tokens. Considering the speed efficiency xAI promises, the model is easy on the budget. But that could also mean that it would burn out tokens fast when it comes to a complex task, something developers should consider.
Moreover, xAI’s affirms that the model is a “versatile choice” for common coding tasks.
Its “strength lies in delivering strong performance in an economical, compact form factor, making it a versatile choice for tackling common coding tasks quickly and cost-effectively.”
While Musk’s attempt with Grok Code Fast 1 is aimed at speed, this isn’t the first time tech giants are exploring dedicated coding agents. Major companies including Microsoft had already made debut on making AI coding assistants available to users.
In May, Mustafa Suleyman led Microsoft AI also introduced the GitHub Copilot feature – a coding agent at the annual Build software developer conference. Similarly, Sam Altman led OpenAI also made its agent Codex, available to ChatGPT Plus users following that June.
