Billionaire Elon Musk is currently involved in a countersuit filed by OpenAI. Despite this legal challenge, his AI company, xAI, has decided to make its flagship Grok 3 model accessible via an API.
In the past few months, xAI introduced Grok 3, a direct competitor to models like OpenAI’s GPT-4o and Google’s Gemini. Grok 3 possesses the capability to analyze images and answer inquiries, and it is integrated into various features on Musk’s social network, X, which was notably acquired by xAI in March.
xAI offers two versions of its flagship model through its API: Grok 3 and Grok 3 Mini, both equipped with reasoning capabilities.
The pricing for Grok 3 is set at $3 per million tokens (approximately 750,000 words) for input into the model and $15 per million tokens generated by the model. Grok 3 Mini is priced at $0.30 per million input tokens and $0.50 per million output tokens.
Faster versions of Grok 3 and Grok 3 Mini are available at higher prices: $5 per million input tokens and $25 per million output tokens for Grok 3, while Grok 3 Mini will cost $0.60 per million input tokens and $4 per million output tokens.
The cost of Grok 3 is relatively high compared to its competitors. xAI aligns its pricing with Anthropic’s Claude 3.7 Sonnet, which also features reasoning capabilities, but Grok 3 is more costly than Google’s recently launched Gemini 2.5 Pro. Gemini achieves higher scores than Grok 3 in popular AI benchmarks. Previously, xAI faced allegations of misleading the public regarding Grok 3’s benchmark results.
Some users on the platform X have highlighted that Grok 3, accessible through xAI’s API, has a smaller context window than what the model supposedly supports. “Context window” refers to the number of tokens the model can process at once. The API limits the context window to 131,072 tokens, or roughly 97,500 words, which falls short of the 1 million tokens xAI claimed Grok 3 could handle in late February.
When Musk initially announced Grok about two years ago, he described the AI model as edgy, unfiltered, and counter to the “woke” narrative, asserting it would address controversial questions that other AI systems typically avoid. Grok and its predecessor, Grok 2, largely fulfilled this promise, providing responses with colorful language not commonly found in models like ChatGPT.
However, prior versions of Grok hedged on political topics and avoided crossing certain boundaries. A study found that Grok demonstrated a political lean to the left on issues such as transgender rights, diversity programs, and inequality.
Musk attributed this behavior to Grok’s training data, derived from public web pages, and he has committed to adjusting Grok to be more politically neutral. Despite incidents like the brief censorship of unflattering mentions of President Donald Trump and Musk, it remains unclear if xAI has achieved political neutrality at the model level or the potential long-term implications of such an adjustment.