SpaceXAI Grok 4.6 Launches: GPT-5.6 Performance at Half the Price

The model, an incremental update following SpaceX's acquisition of xAI, is designed to maintain focus across extended, multi-step tasks such as software engineering, technical research, and automated system operations.

Third-party testing from Artificial Analysis gives Grok 4.6 a score of 61 on its Intelligence Index, tying OpenAI's GPT-5.6 Sol Max and surpassing Moonshot AI's Kimi K3. It ranks third globally behind Anthropic's Claude Opus 5 and Fable 5 models.

SpaceXAI focused on solving operational drift in long-running AI workloads through extended post-training routines using curated synthetic reasoning, technical data, and targeted reinforcement learning across web development and system design. Internal evaluations show the system performs more frequent self-verification checks before completing actions, improving first-try success rates on multi-stage software projects.

Standard API pricing starts at $2 per million input tokens and $6 per million output tokens for prompts under 200,000 tokens, undercutting comparable flagship models by more than half. Prompts exceeding that threshold incur double rates of $4 per million input and $12 per million output across the full request.

Grok 4.6 is available immediately through the $30 monthly SuperGrok plan in Grok Build and via the Cursor code editor, which SpaceX recently acquired. Cloudflare, OpenRouter, and Vercel also host API access. However, adoption faces non-technical hurdles: the Grok brand has a history of safety and content governance controversies, and regulators in the UK and EU continue investigating historical data handling and platform safety controls.