SpaceXai Launches Grok 4.5: The New AI Player Shakes Up the Market

Grok 4.5, SpaceXAI's latest release, comes at a time when competition among AI models is reaching its peak. The launch on July 8 had pricing of $2 per million input tokens and $6 per million output tokens, making it an attractive option compared to similar models from competitors such as Anthropic and OpenAI. The price reduction of more than 50% compared to established players could well have a significant impact on the choices of companies and developers.

Elon Musk emphasized on social media that Grok 4.5 is “roughly comparable” to Anthropic’s Opus 4.7, but that it is much faster. This claim is particularly relevant for those who prioritize efficiency over pure performance. Musk questioned whether the focus on speed and cost came at the expense of overall performance, but also indicated that real-world use, including by engineering teams at Tesla and SpaceX, validates the model due to its practical applicability.

At the launch, SpaceXAI published four benchmark results that put Grok 4.5's performance in perspective. In the DeepSWE 1.1 test, which measures an AI's capacity to solve software problems, Grok 4.5 achieved a score of 53%, leaving it behind Claude Opus 4.8 (59%) and GPT 5.5 (67%). The result makes it clear that competitive scores are possible, but not at the highest level. However, Grok 4.5 excelled in other benchmarks, such as SWE Bench Pro, where it showed an acceptable 64.7%, offering an advantage over GPT 5.5's 58.6%.

The argument for Grok 4.5 is not just about raw power; it is certainly about efficiency as well. Benchmark results show that Grok 4.5 used an average of 15,954 output tokens for its tasks, while Opus 4.8 consumed an astonishing 67,020 tokens for the same tasks. This means that Grok 4.5 is not only cheaper to use but also enables users to perform more iterations. For investors, this is a crucial point: with lower costs and higher efficiency, the model can be more advantageous in certain cases; however, the lower scores in the quality benchmarks may be a barrier for some.

The ability to make Grok 4.5 available via an API, with support for a context of half a million tokens, opens the door to larger projects and a wider range of applications. This offers particular added value for developers who perform frequently recurring coding tasks. The combination of training with real-world data, such as debugging traces and code changes, strengthens the practical applicability of the model.

Frequently Asked Questions

How does Grok 4.5 compare to other AI models?
Grok 4.5 offers significant cost advantages but lagged behind in benchmark tests compared to models such as Opus 4.8 and Claude Fable 5. This placed it in a unique position within the competitive AI market, where price is often a decisive factor.

What are the main benefits of Grok 4.5 for developers?
The model structure and pricing make it attractive for developers who need to work efficiently with a high volume of coding tasks. The ability to experiment more at lower costs can lead to innovations in the development process.

When is Grok 4.5 available to European users?
SpaceXAI has announced that Grok 4.5 will be available to users in the European Union in mid-July, allowing them to fully benefit from the model's functionalities.