Source : INDIA TODAY NEWS
The AI race has intensified in recent months as Anthropic and OpenAI continue to make advancements when it comes to coding tools. However, Elon Musk’s SpaceXAI (previously xAI) had been trailing behind the two AI labs when it came to coding or other tasks such as finance and legal services with its own AI model, Grok. Now, SpaceXAI has unveiled Grok 4.5 – its most advanced AI model – built with the AI coding platform Cursor.
advertisement
As per SpaceXAI, Grok 4.5 is its strongest AI model ever, and built to handle a variety of things – be it coding, agentic tasks or knowledge work. The company says that Grok 4.5 can also be used for business work, such as finance or legal services, something also offered by Anthropic with Claude.
According to tech trillionaire Elon Musk, Grok 4.5 can match Anthropic’s Claude Opus 4.7 model, but it is “much faster.” He added, “The combination of capability, faster speed and lower cost is what makes it competitive.”
Do note that Grok 4.5 is the first joint model from SpaceXAI and Cursor, and its launch comes weeks after SpaceX formally agreed to acquire Cursor in a deal valuing the startup at $60 billion.
What can Grok 4.5 do?
In a blog post on Wednesday, SpaceXAI said Grok 4.5 was designed to ‘handle difficult, long-running tasks’, especially in software engineering, while also extending to legal and financial work.
The company used the launch to show examples of one-prompt app building, including a Three.js solar system simulation with adjustable time and realistic motion. It also said Grok 4.5 is now the default model in Grok Build and can create complex Excel models using web research and multi-sheet formulas, leave notes for later reference, build diagrams in PowerPoint and write clear prose in Word.
SpaceXAI also shared benchmark scores for Grok 4.5. In AI coding benchmark DeepSWE 1.0, the model outperformed Anthropic’s Opus 4.8, and stood right behind OpenAI’s GPT-5.5. Grok 4.5 was trained across tens of thousands of highly-advanced Nvidia GB300 GPUs.
Grok 4.5 can find cybersecurity flaws
The SpaceXAI model also comes with increased cybersecurity capabilities. Cursor said it had taken steps to “detect and block bad actors” while seeking to “preserve legitimate security work, including finding and patching vulnerabilities, while restricting the workflows most likely to cause harm.”
This comes at a time when the US government had previously limited access to modliels like Claude Fable 5 and GPT-5.6 due to concerns over cybersecurity capabilities. Though the US has allowed the general release for both models since then.
The release is part of a broader effort by Musk’s company to catch up in the AI race and win more business customers. Elon Musk had said earlier this year that his AI startup, known as xAI before it merged with SpaceX, had fallen behind on coding, leading to staffing changes.
SpaceXAI released its first coding agent in May to compete with Anthropic’s products. Grok 4.5 is now available in Grok Build, in Cursor on all plans and through the SpaceXAI console, though it is not yet available in the European Union.
The company has priced it at $2 (roughly Rs 190) per million input tokens and $6 (roughly Rs 570) per million output tokens, lower than Claude Opus 4.8 at $5 (roughly Rs 480) and $25 (roughly Rs 2390) respectively, while OpenAI’s GPT 5.6 Terra is priced at $2.5 (roughly Rs 240), and $15 (roughly Rs 1430) respectively.
– Ends
SOURCE :- TIMES OF INDIA
