3 Sources
[1]
Elon Musk says Grok 4.5 is now in private beta, claims performance near or above Anthropic's Opus
Musk said reinforcement learning is continuing to improve the model and that "Grok Build," xAI's coding-agent harness, is being refined daily. He framed the SpaceX and Tesla deployment as an internal proving ground before any wider release. Elon Musk on Saturday announced that xAI's newest model, Grok 4.5, has entered private beta testing in SpaceX and Tesla, and that early internal evaluations show the model performing close to, and in some cases surpassing, Anthropic's Opus model. "Grok 4.5, based on our 1.5T V9 foundation model, with Cursor data added in supplemental training, is now in private beta at SpaceX & Tesla. Early evals show performance close to, perhaps exceeding Opus," he wrote on X. The supplemental Cursor data used to train V9 was from developer workflows, multi-file edits, refactors, and iterative correction cycles, which xAI has positioned as a way to sharpen the model's coding performance. Cursor is the AI coding assistant SpaceX agreed to acquire earlier in June. Musk said reinforcement learning is continuing to improve the model and that "Grok Build," xAI's coding-agent harness, is being refined daily. He framed the SpaceX and Tesla deployment as an internal proving ground before any wider release. Musk said SpaceX will ship a completely new model, trained from scratch, every month for the rest of the year. Grok 4.5 marks the latest entry in a model line that's been hard to track even for close observers. xAI's V9-Medium model -- the 1.5 trillion parameter, Cursor-trained system -- was first announced by Musk in late May, with some reports speculating it would ultimately ship as "Grok 5," while others expected it to land as an interim 4.x release. The "close to, perhaps exceeding Opus" claim comes from Musk's own account of internal evaluations and no third-party benchmark results have been published. The Cursor deal SpaceX will buy the startup behind the popular AI coding agent Cursor, Anysphere, for $60 billion in an all-stock deal to boost its presence in the lucrative enterprise AI tools market. The deal gives xAI, acquired by SpaceX in February, a stronger hold in coding, one of the first areas that have helped companies turn AI into a source of revenue from businesses. Cursor is among several Silicon Valley startups that have helped developers use AI to automate coding, making it a key rival to market leaders Anthropic and OpenAI. But a lack of access to computing power has hampered Cursor's growth. SpaceX had been eyeing Cursor for months and in April unveiled an option to either buy the startup for $60 billion later this year, or pay $10 billion for a partnership.
[2]
Elon Musk Unveils Grok 4.5 Beta, Claims it Beats Claude Opus AI
Elon Musk's xAI has launched Grok 4.5 in private beta, claiming the new model surpasses Anthropic's Claude Opus in key benchmarks as competition among leading AI companies intensifies. Elon Musk's artificial intelligence company, xAI, has introduced Grok 4.5 in private beta, marking the latest update to its flagship chatbot. The company claims the new model delivers stronger performance than rival AI systems, including Anthropic's Claude Opus, across several benchmarks. The release comes as AI companies continue to roll out increasingly capable models, with firms racing to improve reasoning, coding, and conversational abilities while attracting enterprise customers.
[3]
Musk says Grok 4.5 enters private beta at SpaceX and Tesla By Investing.com
Investing.com -- Elon Musk announced Sunday that Grok 4.5 has entered private beta testing at SpaceX and Tesla. The model is built on a 1.5 trillion parameter V9 foundation model with Cursor data incorporated during supplemental training. Initial evaluations indicate the model's performance approaches or potentially surpasses Opus, according to Musk. He stated that reinforcement learning continues to improve the model substantially, while the Grok Build harness advances daily. SpaceX plans to release completely new models trained from scratch each month throughout this year. This article was generated with the support of AI and reviewed by an editor. For more information see our T&C.
Share
Copy Link
Elon Musk revealed that xAI's Grok 4.5 has entered private beta testing at SpaceX and Tesla, with early evaluations showing performance close to or exceeding Anthropic's Opus. Built on a 1.5 trillion parameter foundation model enhanced with Cursor data, the AI model represents xAI's latest push in enterprise AI tools. SpaceX plans to ship completely new models trained from scratch monthly through year-end.
Elon Musk announced on Saturday that xAI's newest AI model, Grok 4.5, has entered private beta testing at SpaceX and Tesla, marking a significant step in the company's effort to compete with leading AI developers
1
. The model is built on a 1.5 trillion parameter V9 foundation model, with supplemental Cursor data incorporated during training to enhance its capabilities3
. According to Musk's post on X, early internal evaluations show performance near or above Anthropic's Opus model, positioning Grok 4.5 as a potential competitor to one of the industry's leading AI systems1
.
Source: Analytics Insight
The supplemental Cursor data used to train the V9 foundation model came from developer workflows, multi-file edits, refactors, and iterative correction cycles, which xAI has positioned as a strategic way to sharpen the model's coding performance
1
. Cursor is the AI coding assistant that SpaceX agreed to acquire earlier in June through a $60 billion all-stock deal with Anysphere, the startup behind the popular tool1
. The acquisition gives xAI, which SpaceX acquired in February, a stronger foothold in coding—one of the first areas where companies have successfully turned AI into a revenue source from businesses1
.Musk stated that reinforcement learning continues to improve the model substantially, while "Grok Build," xAI's coding-agent harness, is being refined daily
1
. He framed the private beta testing at SpaceX and Tesla as an internal proving ground before any wider release, suggesting the companies will serve as real-world testing environments for the AI model1
. In an ambitious development timeline, Musk announced that SpaceX will ship a completely new model, trained from scratch, every month for the rest of the year .
Source: ET
Related Stories
The claim that xAI's Grok 4.5 model beats Claude Opus AI comes from Musk's own account of internal evaluations, and no third-party benchmark results have been published to verify these assertions
1
. The release comes as AI companies continue to roll out increasingly capable models, with firms racing to improve reasoning, coding, and conversational abilities while attracting enterprise customers2
. Grok 4.5 marks the latest entry in a model line that's been difficult to track even for close observers, as xAI's V9-Medium model was first announced by Elon Musk in late May, with speculation about whether it would ship as "Grok 5" or an interim 4.x release1
. The $60 billion Cursor acquisition positions xAI to compete more directly with market leaders Anthropic and OpenAI in the lucrative enterprise AI tools market, particularly in coding automation1
.Summarized by
Navi
[1]
[2]
1
Technology

2
Technology

3
Policy and Regulation
