\n<\/p>\n
Billionaire Elon Musk might\u2019ve just been countersued by OpenAI<\/a>. But that isn\u2019t stopping his AI company, xAI, from making its flagship Grok 3 model available via an API<\/a>. <\/p>\n It has been several months since xAI unveiled Grok 3, the company\u2019s answer to models like OpenAI\u2019s\u00a0GPT-4o<\/a>\u00a0and Google\u2019s\u00a0Gemini<\/a>. \u00a0Grok 3 can analyze images and respond to questions, and powers a number of features on Musk\u2019s social network, X, which not-so-coincidentally acquired xAI in March.<\/p>\n xAI is offering two flavors of Grok 3 in its API: Grok 3 and Grok 3 mini with \u201creasoning\u201d capabilities. Grok 3 is priced at $3 per million tokens (~750,000 words) fed into the model and $15 per million tokens generated by the model. As for Grok 3 mini, it\u2019s $0.30 per million input tokens and $.50 per million output tokens. Speedier versions of Grok 3 and Grok 3 mini are available at a significant premium.<\/p>\n Grok 3 isn\u2019t exactly cheap relative to the competition. It matches the pricing of Anthropic\u2019s Claude 3.7 Sonnet, which also offers reasoning capabilities, and it\u2019s more expensive than Google\u2019s recently released Gemini 2.5 Pro, which achieves generally higher scores than Grok 3 across popular AI benchmarks. (Not for nothing, xAI has been accused of being misleading<\/a> in its Grok 3 benchmark reporting.)<\/p>\n As several users on X pointed out, Grok 3 via xAI\u2019s API also has a smaller context window than the model is allegedly capable of supporting. \u201cContext window\u201d refers to how many tokens the model can process in a single go. The API maxes out at 131,072 tokens, or around 97,500 words \u2014 short of the 1 million tokens (equivalent to roughly 750,000 words) xAI claimed that Grok 3 supported in late February. <\/p>\n When Musk announced Grok roughly two years ago, he pitched the AI model as edgy, unfiltered, and anti-\u201cwoke\u201d \u2014 in general, willing to answer controversial questions other AI systems won\u2019t. He delivered on some of that promise. Told to be vulgar, for example, Grok and Grok 2 would happily oblige, spewing colorful language you likely wouldn\u2019t hear from\u00a0ChatGPT<\/a>.<\/p>\n But Grok models prior to Grok 3\u00a0hedged<\/a>\u00a0on political subjects and wouldn\u2019t cross\u00a0certain boundaries<\/a>. In fact,\u00a0one study<\/a>\u00a0found that Grok leaned to the political left on topics like transgender rights, diversity programs, and inequality.<\/p>\n Musk has blamed the behavior on Grok\u2019s training data \u2014 public web pages \u2014 and\u00a0pledged<\/a>\u00a0to \u201cshift Grok closer to politically neutral.\u201d It\u2019s not yet clear whether xAI has achieved that goal, and what the consequences might be.<\/p>\n<\/div>\n