SpaceXAI unveils Grok 4.6 AI model, matching GPT-5.6 Sol and Claude Fable 5

SpaceXAI has introduced Grok 4.6, a new flagship AI model that the company says is comparable in capabilities to OpenAI's GPT-5.6 Sol and Anthropic's Claude Fable 5. Designed for 'long-lived' agents, Grok 4.6 can autonomously work on tasks over extended periods, restarting itself and progressing towards goals, handling unfamiliar codebases, and testing its own results. In the Artificial Analysis Intelligence Index, Grok 4.6 scored 61 points, surpassing the open-weights Chinese model Kimi K3 from Moonshot AI, and matching GPT-5.6 Sol's top reasoning level, trailing Fable 5 Max by one point. Compared to Grok 4.5, the new version shows improvements across all listed benchmarks including CursorBench, FrontierCode, APEX-Agents, and Terminal-Bench, thanks to a longer training cycle, better engineering data, and expanded reinforcement learning for coding and knowledge work. The model also demonstrates significant gains in code development, terminal operations, information handling, and agent performance while keeping API pricing at $2 per million input tokens and $6 per million output tokens, positioning it as a mid-priced option among leading proprietary and open models. Grok 4.6 is available now in Cursor, Grok Build, and via APIs from SpaceXAI, OpenRouter, Vercel, and Cloudflare.