Overview
- xAI publicly released Grok 4.5 on July 8–9 with a 500,000‑token context window and direct distribution through Cursor, Grok Build and the xAI API.
- xAI published pricing at $2 per million input tokens and $6 per million output tokens and billed the model as highly token efficient for coding and agent workflows.
- Early industry reports praised Grok’s agent and coding performance while independent benchmarks placed it lower on overall and coding scores, creating a split between company‑backed and third‑party results.
- A leaked Tesla memo urging engineers to use Grok when possible and capping third‑party AI spending prompted Elon Musk to clarify that staff should try Grok but keep using better tools when those outperform it.
- Reports that Grok runs on tens of thousands of NVIDIA GB300 GPUs have analysts tracking how lower per‑task compute needs and higher adoption could squeeze decentralized GPU rental markets and prompt competitive price or capability responses from rivals.