xAI released Grok 4.5 on July 8, achieving a 29.0% resolution rate on the SWE Marathon benchmark, surpassing Anthropic’s Claude Opus 4.8 at 26.0% and Fable at 24.0%. The model was trained on tens of thousands of NVIDIA GB300 GPUs with reinforcement learning for software engineering tasks, offering inference speed of roughly 80 transactions per second. xAI claims up to 4.2 times better token efficiency compared to leading competitors, with pricing at $2 per million input tokens and $6 per million output tokens. Grok 4.5 launched via the Grok app, xAI console, and API, with EU availability expected by mid-July, and also posted strong results on DeepSWE and Terminal Bench 2.1. No independent third-party verification of the SWE Marathon results has occurred. The article notes that Grok 4.5 has nothing to do with crypto, but its downstream effects on coding agents could impact areas like Solidity audits and decentralized compute networks.