SpaceXAI releases Grok 4.6, claiming GPT
Key Points:
- SpaceXAI has released Grok 4.6, claiming its intelligence rivals OpenAI’s GPT-5.6 Sol and Anthropic’s Claude Fable 5, aiming to enhance Grok’s reputation and capabilities.
- Grok 4.6 is optimized for complex tasks such as long-running agents, coding, knowledge work, and advanced visual projects, with improved ability to stay on task and self-verify outputs.
- On the Artificial Analysis Intelligence Index, Grok 4.6 scored 61, matching GPT-5.6 Sol at its highest reasoning level and narrowly trailing Claude Fable 5 by one point.
- The improvements over Grok 4.5 stem from extended supplemental training, enhanced engineering data, and increased reinforcement learning focused on coding and knowledge work.
- Grok 4.6 is available through multiple platforms including Cursor and SpaceXAI API, with competitive pricing starting at $2 per million input tokens and promotional doubled usage for Cursor and Grok Build users in the first week.