Grok 4.5: Cheaper AI for Coding & Knowledge Work

SpaceXAI has launched Grok 4.5, which it calls its smartest model for coding, agentic tasks, and knowledge work. The company aims for it to be an "Opus-class" AI, directly comparing it to Anthropic's high-end Claude Opus family. What's interesting is the price. Grok 4.5 starts at $2 per million input tokens and $6 per million output tokens. SpaceXAI says this model is designed for work, not small talk, and can handle complex tasks while using fewer tokens. This could make it cheaper to run at scale. The company launched Grok 4.5 on July 8, 2026, after training it alongside Cursor, an AI coding platform. SpaceXAI claims Grok 4.5 beats some comparable models on several software-engineering tests. The bottom line is that cheaper, powerful AI agents could significantly impact teams, especially those in regions where serious AI tools often bill in dollars.