Elon Musk Buries Beef With Frenemy, Calls Rival Obvious AI Leader

Just like Marvel screen hero Tony Stark - who is loosely based on Elon Musk - the real-life boss of SpaceXAI and Tesla never ceases to surprise. His public spat with friend-turned-foe Sam Altman of OpenAI is well known. Anthropic too has been on the receiving end - Musk has ranted publicly many times saying things like the company's models "hate Whites & Asians, especially Chinese, heterosexuals and men". He even labeled the company "misanthropic and evil" and accused it of "stealing training data at massive scale."

In a change of heart, Musk on Thursday admitted he was wrong about Anthropic, even calling them the obvious leader in AI.

"I was clearly wrong about Anthropic. They are obviously currently the leader in AI. No company has released a model as good as Mythos/Fable and they will undoubtedly have Mythos 2 ready soon," Musk said on X, in response to user who was dissing Anthropic while glorifying Musk's SpaceXAI.

Also read: Mythos A Weapon, Needs Gun License, Testers Warned Anthropic

The user who goes by kache on X had said that SpaceXAI now has a legitimate frontier model that competes with Opus 4.8. "Also, Anthropic is completely reliant on the compute rented from SpaceXAI. If Elon wanted to kill Anthropic, he could. Iirc the compute lease was short term, 6 months from May without renewal promise."

Musk is known to often engage with users on X (also owned by him). He responded "I would never cut them (Anthropic) off in a way that hurt them badly, even as a competitor. That's not my style."

Highlighting his ethics and integrity Musk went on to add:  "Tesla open sourced its patents and we made the Supercharger network available to all competitors, even though we could have made it a walled garden." 

"SpaceX launches competing satellite systems with no increase in price or use of unfair terms. Even my worst enemies can attack me on this platform."

Interestingly the same user kache last year had posted on X that: "Anthropic is going to lose (the AI race)," to which Musk has snidely replied: "Winning was never in the set of possible outcomes from Anthropic."

So Thursday's statement is, of course, a huge U-turn from that. 

The exchange also highlighted an unusual dynamic in the AI race. Although SpaceXAI and Anthropic compete head-to-head in frontier AI, the two companies also have a business relationship. Earlier this year, Anthropic struck a major deal to lease AI computing capacity from SpaceXAI's data centres, reflecting a growing trend in the industry where rival AI labs depend on one another's infrastructure as demand for GPUs (Graphics Processing Units) continues to outstrip supply. That is the arrangement the X user was referring to when claiming Anthropic was reliant on SpaceXAI's compute. Musk, however, said he had no intention of using that commercial relationship against the company.

The AI Race

SpaceXAI, formerly known as xAI, unveiled its latest model, Grok 4.5 on Thursday. The launch marked the company's first major AI release since it went public last month, a move that made Musk the world's first trillionaire.

According to Musk, the new model is approaching the capabilities of Anthropic's flagship AI systems while offering faster performance and lower operating costs.

"Our internal assessment is that Grok 4.5 is roughly comparable to Opus 4.7, but much faster. The combination of capability, faster speed and lower cost is what makes it competitive," Musk wrote on X, formerly Twitter.

Opus 4.7 is one of Anthropic's advanced AI models positioned closely behind its flagship Fable 5 system, which is powered by the company's most capable AI architecture, Mythos. The model drew global attention recently after the US government banned it for a couple of weeks over security concerns. The model's advanced cybersecurity capabilities could identify vulnerabilities in critical infrastructure that had remained undiscovered for decades.

Musk said the company is prioritising practical performance over benchmark scores.
"We are closing the loop on real-world usefulness, not benchmarks. Hardcore engineers at Tesla & SpaceX find Grok 4.5 genuinely useful, which is what actually matters," he said.

In a blog post announcing the release, SpaceXAI said Grok 4.5 is designed to perform a broad range of knowledge-based tasks that businesses increasingly expect AI to handle. These include software development, application building, research, writing, office administration and other routine professional work.

SpaceXAI has also positioned Grok 4.5 aggressively on pricing. The company said it will charge $2 per million input tokens and $6 per million output tokens, significantly below the rates charged by several competing frontier AI models.

For comparison, Anthropic prices Opus 4.7 at $5 per million input tokens and $25 per million output tokens. OpenAI's premium model, Sol, is priced at $5 per million input tokens and $30 per million output tokens, while its lower-cost Luna model is priced at $1 per million input tokens and $6 per million output tokens.

Whether Grok 4.5 can consistently match the performance of leading models from Anthropic and OpenAI at those prices, however, will likely become clearer once developers begin testing it more extensively.

The Grok 4.5 launch came during a packed week for the AI industry. OpenAI also rolled out GPT-5.6, its latest flagship model, from Thursday. The release had reportedly been delayed because of security-related reviews by the Trump administration. OpenAI has described GPT-5.6 as its most capable AI model to date.
Mark Zuckerberg's Meta on Thursday also jumped in to the AI coding market to chase OpenAI, Anthropic. It debuted Muse Spark 1.1 with a preview available to developers.