xAI CEO Elon Musk made an unexpected statement, publicly acknowledging that his flagship neural network Grok 4.5 is inferior to Anthropic's Claude Fable 5 model in overall quality. However, as Musk emphasizes, this does not mean a complete defeat — Grok 4.5 remains one of the fastest and most cost-effective models on the market.
In a post on platform X, Musk responded to a user who claimed that Grok 4.5 performs at the level of Fable but costs significantly less. The billionaire clarified that while Fable indeed surpasses his model in several parameters, such a high level is simply excessive for most everyday tasks. This admission is a rare case for the head of a major technology company, who usually avoids public concessions to competitors.
Comparative Analysis: Grok 4.5 vs. Competitors
Independent tests confirm that SpaceXAI's strategy focuses on affordability and speed. The cost of Grok 4.5 is only $2 per million input tokens and $6 per million output tokens. For comparison: Claude Opus 4.8 costs $5 and $25 respectively, while GPT-5.6 Sol costs $5 and $30. The price difference reaches 5–10 times, making Grok 4.5 an attractive solution for mass adoption.
In the Artificial Analysis intelligence index, the model ranked fourth with 54 points, behind Fable 5 (60 points), Opus 4.8 (56), and GPT-5.5 (55). However, results vary in applied tests. On SWE Marathon, Grok 4.5 showed the best result — 29%, outperforming all competitors. Meanwhile, on SWE Bench Pro, the model scored 64.7%, trailing behind Claude Opus 4.8 (69.2%) and Claude Fable 5 (80.4%).
Particular attention is drawn to the AutomationBench-AA test, where Grok 4.5 scored 51.4%, ahead of Fable 5 (48.6%) and Opus 4.8 (48.5%). At the same time, the cost of solving one task with Grok was only $0.34 compared to $1.35 for Fable and $1.46 for Opus. Musk previously used this performance metric as the main proof of his network's strength.
The Flip Side of Efficiency
However, such cost-effectiveness also has drawbacks. Experts note a higher frequency of rule violations in Grok 4.5, which creates additional risks for corporate deployment of AI agents. The model can generate unpredictable responses, which is critical for business environments requiring strict protocol adherence.
The main point of this release boils down to simple economics: developers show that a strong price difference outweighs a slight lag in tests. For the mass market, where speed and cost matter, Grok 4.5 remains a competitive product. The only question is how long such a strategy will work in the rapidly evolving AI industry.
My analysis: Musk's public admission is not weakness but a calculated marketing move. He shifts the focus from the race for absolute quality to practical value, which could attract the small and medium business segment for whom price is decisive. However, in the long term, xAI needs to close the gap with leaders, otherwise the model risks remaining a niche product.