SpaceXAI launches Grok 4.5, an AI model focused on efficiency and cost

SpaceXAI introduced Grok 4.5 this Wednesday, the 8th, representing a new generation of its artificial intelligence model. This new system is designed to perform various tasks, including programming, application development, research, text composition, and administrative support.

Model Performance and Efficiency

The company claims that the updated version combines high performance with significant optimization in token consumption, the metric used to measure data processing by AI models. According to the company, this system demonstrates twice the efficiency observed in other market-leading models.

This launch occurs just weeks after SpaceXAI's initial public offering and places Grok 4.5 in a fiercely competitive landscape among major artificial intelligence developers, a period characterized by the continuous introduction of new competing models.

Competitive Positioning and Costs

According to SpaceXAI, Grok 4.5 was conceived as a comprehensive productivity tool, capable of managing various types of intellectual work that technology companies seek to automate. Performance test results released by the company itself suggest that the model approaches the most prominent systems currently available, although it does not yet hold absolute leadership in existing rankings.

Elon Musk, founder of SpaceXAI, drew an analogy between the launch and Opus, Anthropic's model aimed at complex tasks. In a post on SpaceXAI's X network, Musk stated that Grok 4.5 possesses capabilities comparable to the Opus category but offers additional benefits in terms of speed, efficiency, and price.

In another communication, the executive assessed that the new system would achieve performance close to Opus 4.7, but with more agile operation. For Musk, the combination of technical competence, operational speed, and lower cost constitutes the product's main distinguishing factor.

Pricing Structure and Comparisons

The company established that Grok 4.5 will be available for $2 per million input tokens and $6 per million output tokens. This pricing structure aims to demonstrate an economic advantage over competing models.

For comparison, the article mentions that Opus 4.7 costs $5 per million input tokens and $25 per million output tokens. OpenAI operates with different values depending on the model version: Sol, its most expensive option, costs $5 per million input tokens and $30 per million output tokens, while Luna costs $1 per million input tokens and $6 per million output tokens.

The launch of Grok 4.5 coincides with a week of great activity in the sector. OpenAI also plans to introduce GPT 5.6, which the company itself describes as its most advanced model to date. As reported, the availability of this version had previously been restricted by the Trump administration due to security concerns.