SpaceXAI Launches Grok 4.5: High-Speed Coding Agent to Rival Opus
SpaceXAI has launched Grok 4.5, a high-performance reasoning model optimized for coding and agentic tasks. Trained on GB300 GPUs, Grok 4.5 runs at 80 tokens per second with improved token efficiency. Billed at two dollars per million inputs and six dollars per million outputs, the model integrates with Cursor and Grok Build.







