SpaceX AI launches Grok 4.6 to challenge leading AI models
SpaceX AI just launched Grok 4.6, its newest AI model designed for advanced coding and knowledge work.
The company says Grok 4.6 can now match OpenAI's GPT-5.6 Sol in reasoning and stands toe to toe with Anthropic's Claude Fable 5; pretty bold claims!
This follows its recent Cursor acquisition and the rollout of Grok Bot, showing SpaceX AI is serious about challenging the biggest names in AI.
Grok 4.6 trained for software engineering
Grok 4.6 has been trained on a ton of carefully chosen data, making it better at handling tricky multistep tasks, understanding fresh codebases, and checking its own work.
It also got extra training for things like software engineering and kernel optimization, helping it score a solid 61 on the Artificial Analysis Intelligence Index, right up there with GPT-5.6 Sol and beating out its own previous version in key benchmarks.
You can now use Grok 4.6 on platforms like Cursor, Grok Build, OpenRouter, and Vercel starting at $2 per 1 million input tokens.
With improved security features for things like vulnerability patching and design work, this launch is a big step forward for SpaceX AI as it aims to become a major player in the field.