Elon Musk's xAI has released Grok 4.5, a language model aimed at coding tasks, and framed the launch around cost efficiency. xAI published pricing that places Grok 4.5 at $2 per million input tokens and $6 per million output tokens.
Performance positioning and comparisons
The company claims Grok 4.5 is close to Opus 4.7 in quality. In promotional material and statements, Elon Musk compared the model's speed and price‑performance primarily against Opus and OpenAI's GPT family.
Competitors omitted from the comparison
Notably, several other large language model providers—particularly some Chinese labs—were not included in the comparisons. The coverage points to firms such as DeepSeek that reportedly offer substantially lower prices than the figures xAI publicized, and in some cases distribute model weights freely. Omitting these players affects how meaningful Grok 4.5's price advantage appears in a global market context.
Why this matters
Which competitors are chosen for benchmarking affects market perception. If a provider compares itself only to rivals it can undercut on price or speed, the resulting message emphasizes relative positioning rather than a true breakthrough. The $2/$6 pricing may undercut many Western frontier models that begin in the tens of dollars, but it is less definitive if competitors exist that already price below those levels or offer free access to model weights.
Summary
xAI promoted Grok 4.5 as a low‑cost, high‑performance coding model with published token prices of $2 input and $6 output per million, and compared it against Opus 4.7 and GPT models. However, the decision not to mention Chinese LLM providers such as DeepSeek, which reportedly charge much less or release weights freely, means the claimed cost advantage may not represent a global market breakthrough.
