SpaceXAI releases Grok 4.5, which Elon describes as an ‘Opus-class model’

SpaceXAI has officially launched its latest artificial intelligence model, Grok 4.5, marking the first major release since the company went public a few weeks ago. In a detailed blog post released on Wednesday, SpaceXAI described Grok 4.5 as a robust solution capable of addressing a wide range of tasks that the AI sector has aimed to automate, including coding, app development, administrative work, research, and various forms of routine intellectual tasks. The company claims that Grok 4.5 offers impressive token efficiency, boasting “twice the token efficiency” compared to leading models in the market. If proven accurate in practical applications, this could provide a significant edge to SpaceXAI, especially as the cost of tokens continues to be a pressing concern for AI users. On the same day, the company shared benchmark metrics that suggested Grok's competitive standing against other top models, although it fell just short of being the best-in-class. Founder Elon Musk took to his social media platform, X, to draw a comparison between Grok 4.5 and Opus, a language model from competitor Anthropic designed for high-stakes tasks. He stated, “Based on strong positive feedback from customers in our beta test program, @SpaceXAI will make Grok 4.5 available to the public tomorrow. It is an Opus-class model but faster, more token-efficient, and lower cost.” Musk further elaborated, indicating that internal assessments suggest Grok 4.5 is comparable to Opus 4.7 but operates at a significantly faster pace, highlighting its blend of capability, speed, and affordability. The pricing for Grok 4.5 is also noteworthy, set at $2 per million input tokens and $6 per million output tokens, which positions it competitively against other models. For context, Opus 4.7 costs $5 for each million input tokens and $25 for output tokens. Additionally, OpenAI has a tiered pricing structure for its models, with its premium model, Sol, priced at $5 for 1 million input tokens and $30 for 1 million output tokens, while its more economical option, Luna, costs $1 for 1 million input and $6 for output. This week is particularly significant for AI advancements, as OpenAI is preparing to unveil GPT 5.6, touted as its most powerful model to date. This release, which had been previously delayed due to security concerns from the Trump administration, is being anticipated as the “strongest model yet.”