xAI Pushes Forward with Grok 3 API Release Amid OpenAI Legal Drama
Legal trouble or not, Elon Musk is pressing full steam ahead. While rumors swirl that OpenAI may have filed a countersuit against him, his AI venture, xAI, isn’t slowing down. In fact, they’ve just opened API access to their flagship model, Grok 3, signaling a major move in the battle for AI dominance.
Grok 3 vs The Giants: What Makes It Stand Out?
Launched a few months ago as a challenger to OpenAI’s GPT-4o and Google’s Gemini, Grok 3 is designed to handle both image analysis and advanced text-based queries. It's already integrated into Musk’s social media platform X, formerly Twitter, which also acquired xAI back in March.
But this latest development isn’t just about flashy features—it’s about accessibility and scale.
Two Versions, One Goal: Smarter Reasoning at Scale
Through the new API, developers can now tap into two variants of the model:
-
Grok 3
-
Grok 3 Mini
Both models come equipped with “reasoning” capabilities—meaning they’re built to think through complex problems, not just regurgitate facts.
Here’s how the pricing breaks down:
| Model | Input (per 1M tokens) | Output (per 1M tokens) |
|---|---|---|
| Grok 3 | $3 | $15 |
| Grok 3 Mini | $0.30 | $0.50 |
Need more speed? Premium versions are also available:
-
Grok 3 Premium: $5/M input, $25/M output
-
Grok 3 Mini Premium: $0.60/M input, $4/M output
For context, 1 million tokens roughly equals 750,000 words—enough to process entire books in a single pass.
Is Grok 3 Competitive in the Market?
When it comes to cost, Grok 3 is on par with Anthropic’s Claude 3.7 Sonnet, another reasoning-heavy model. But it's notably more expensive than Google’s Gemini 2.5 Pro, which currently outperforms Grok 3 on several popular AI benchmarks.
Adding to the scrutiny, some users have pointed out discrepancies in xAI’s marketing. Although Grok 3 was touted as having a 1 million token context window, its API currently maxes out at 131,072 tokens—around 97,500 words. That’s a far cry from the initially promised capacity.
From day one, Musk pitched Grok as an AI willing to “say the unsayable.” Positioned as a raw, unfiltered alternative to more sanitized models like ChatGPT, Grok and its predecessor, Grok 2, had no problem throwing around colorful language when prompted.
Still, despite its edgy persona, earlier Grok versions shied away from hot-button political topics. A study even revealed that the model leaned left on issues like transgender rights, diversity, and economic inequality—a surprising twist, considering Musk’s vocal criticism of “woke” culture.
Political Leanings, Data Bias, and Ongoing Adjustments
Musk chalked up the model’s leftward leanings to its training data, which was largely pulled from public web sources. In response, he vowed to steer the model toward political neutrality. But apart from a few headline-grabbing blunders—such as the brief censorship of content related to Donald Trump and Musk himself—it’s still unclear if Grok 3 has genuinely reached that middle ground.
What’s certain is that xAI is playing a long game, betting big on transparency, user control, and raw computational power. And with the API now out in the wild, we’ll soon see whether Grok 3 can carve out a lasting place in the rapidly evolving world of generative AI.